Turbulent flow over a NACA 4412 airfoil with an angle of attack AoA = 5◦ was analysed using an incompressible direct numerical simulation (DNS) at chord Reynolds number of Rec = 4 · 105. Snapshots of the flow field were analysed using the method of Spectral Proper Orthogonal Decomposition (SPOD) in frequency domain, in order to extract the dominant coherent structures of the flow. Focus is given to two-dimensional disturbances, known to be most relevant for aeroacoustics. The leading SPOD modes show coherent structures forming a wavepacket, with significant amplitudes in the trailing-edge boundary layer and in the wake. To model coherent structures in the turbulent boundary layer, the optimal harmonic forcing and the associated linear response of the flow were obtained using the singular value decomposition of the linear resolvent operator. The resolvent analysis shows that the leading SPOD modes can be associated to most amplified, linearised flow responses. Furthermore, coherent structures in the wake are modelled as the Kelvin-Helmholtz mode from linear stability theory (LST).
A method using gradient-based optimization is introduced for the design of wing profiles with the aim of natural laminar How, as well as minimum wave drag. The Euler equations of gasdynamics, the laminar boundary-layer equations for compressible flows on infinite swept wings, and the linear parabolized stability equations (PSE) are solved to analyze the evolution of convectively unstable disturbances. Laminar-turbulent transition is assumed to be delayed by minimizing a measure of the disturbance kinetic energy of a chosen disturbance, which is computed using the PSE. The shape gradients of the disturbance kinetic energy are computed based on the solutions of the adjoints of the state equations just named. Numerical tests are carried out to optimize the RAE 2822 airfoil with the aim to delay simultaneously the transition, reduce the pressure drag coefficient, and maintain the coefficients of lift and pitch moments. Constraints are also applied on the geometry. Results show a reduction of the total amplification of a large number of disturbances, which is assumed to represent a delay of the transition in the boundary layer. Because delay of the transition implies reduction of the viscous drag, the present method enables shape optimization to perform viscous drag reduction.
The incompressible Navier-Stokes equations have an exact similarity solution for the flow over an infinite rotating disk giving a laminar boundary layer of constant thickness, also known as the von Kármán flow. It is well known now that there is an absolute instability of the boundary layer which is linked to transition to turbulence, but convective routes are also observed. It is these convective modes that we focus on here. A comparison of three different approaches to investigate the convective, so called Type-I, stationary crossflow instability is presented here. The three approaches consist of local linear stability analysis, direct numerical simulations (DNS) and experiments. The ’shooting method’ was used to compute the local linear stability whereas linear DNS was performed using a spectral-element method for a full annulus of the disk, a quarter and 1/32 of an annulus, each with one roughness element in the computational domain. These correspond to simulating one, four and 32 roughness elements on the full disk surface and in addition a case with randomly-distributed roughnesses was simulated on the full disk. Two different experimental configurations were used for the comparison: i) a clean-disk condition, i.e. unexcited boundary-layer flow; and ii) a rough-disk condition, where 32 roughness elements were placed on the disk surface to excite the Type-I stationary vortices. Comparisons between theory, DNS and experiments with respect to the structure of the stationary vortices are made. The results show excellent agreement between local linear stability analysis and both DNS and experiments for a fixed azimuthal wavenumber (32 roughnesses). This agreement clearly shows that the three approaches capture the same underlying physics of the setup, and lead to an accurate description of the flow. It also verifies the numerical simulations and shows the robustness of experimental measurements of the flow case. The effects of the azimuthal domain size in the DNS and superposition of multiple azimuthal wavenumbers in the DNS and experiments are discussed.
Numerical simulations of the flow developing on the surface of a rotating disk are presented based on the linearized incompressible Navier-Stokes equations. The boundary-layer flow is perturbed by an impulsive disturbance within a linear global framework, and the effect of downstream turbulence is modelled by a damping region further downstream. In addition to the outward-travelling modes, inward-travelling disturbances excited at the radial end of the simulated linear region, r(end), by the modelled turbulence are included within the simulations, potentially allowing absolute instability to develop. During early times the flow shows traditional convective behaviour, with the total energy slowly decaying in time. However, after the disturbances have reached r(end), the energy evolution reaches a turning point and, if the location of r(end) is at a Reynolds number larger than approximately R = 594 (radius non-dimensionalized by root v/Omega*, where v is the kinematic viscosity and Omega* is the rotation rate of the disk), there will be global temporal growth. The global frequency and mode shape are clearly imposed by the conditions at r(end). Our results suggest that the linearized Ginzburg-Landau model by Healey (J. Fluid Mech., vol. 663, 2010, pp. 148-159) captures the (linear) physics of the developing rotating-disk flow, showing that there is linear global instability provided the Reynolds number of r(end) is sufficiently larger than the critical Reynolds number for the onset of absolute instability.
This paper considers the analysis and control of fluid flows using tools from dynamical systems and control theory. The employed tools are derived from the spectral analysis of various linear operators associated with the Navier-Stokes equations. Spectral decomposition of the linearized Navier-Stokes operator, the Koopman operator, the spatial correlation operator and the Hankel operator provide a means to gain physical insight into the dynamics of complex flows and enables the construction of low-dimensional models suitable for control design. Since the discretization of the Navier-Stokes equations often leads to very large-scale dynamical systems, matrix-free and in some cases iterative techniques have to be employed to solve the eigenvalue problem. The common theme of the numerical algorithms is the use of direct numerical simulations. The theory and the algorithms are exemplified on flow over a flat plate and a jet in crossflow, as prototypes for the laminar-turbulent transition and three-dimensional vortex shedding.
Many fluid flows, such as bluff body wakes, exhibit stable self-sustained oscillations for a wide range of parameters. Here we study the effect of weak noise on such flows. In the presence of noise, a flow with self-sustained oscillations is characterized not only by its period, but also by the quality factor. This measure gives an estimation of the number of oscillations over which periodicity is maintained. Using a recent theory [P. Gaspard, J. Stat. Phys. 106, 57 (2002)], we report on two observations. First, for weak noise the quality factor can be approximated using a linear Floquet analysis of the deterministic system; its size is inversely proportional to the inner-product between first direct and adjoint Floquet vectors. Second, the quality factor can readily be observed from the spectrum of evolution operators. This has consequences for Koopman/Dynamic mode decomposition analyses, which extract coherent structures associated with different frequencies from numerical or experimental flows. In particular, the presence of noise induces a damping on the eigenvalues, which increases quadratically with the frequency and linearly with the noise amplitude. (C) 2014 AIP Publishing LLC.
The Koopman operator provides a powerful way of analysing nonlinear flow dynamics using linear techniques. The operator defines how observables evolve in time along a nonlinear flow trajectory. In this paper, we perform a Koopman analysis of the first Hopf bifurcation of the flow past a circular cylinder. First, we decompose the flow into a sequence of Koopman modes, where each mode evolves in time with one single frequency/growth rate and amplitude/phase, corresponding to the complex eigenvalues and eigenfunctions of the Koopman operator, respectively. The analytical construction of these modes shows how the amplitudes and phases of nonlinear global modes oscillating with the vortex shedding frequency or its harmonics evolve as the flow develops and later sustains self-excited oscillations. Second, we compute the dynamic modes using the dynamic mode decomposition (DMD) algorithm, which fits a linear combination of exponential terms to a sequence of snapshots spaced equally in time. It is shown that under certain conditions the DMD algorithm approximates Koopman modes, and hence provides a viable method to decompose the flow into saturated and transient oscillatory modes. Finally, the relevance of the analysis to frequency selection, global modes and shift modes is discussed.
This review gives an account of recent research efforts to use feedback control for the delay of laminar-turbulent transition in wall-bounded shear flows. The emphasis is on reducing the growth of small-amplitude disturbances in the boundary layer using numerical simulations and a linear control approach. Starting with the application of classical control theory to two-dimensional perturbations developing in spatially invariant flows, flow control based on control theory has progressed towards more realistic three-dimensional, spatially inhomogeneous flow configurations with localized sensing/actuation. The development of low-dimensional models of the Navier-Stokes equations has played a key role in this progress. Moreover, shortcomings and future challenges, as well as recent experimental advances in this multi-disciplinary field, are discussed.
Elastic filamentous structures found on swimming and flying organisms are versatile in function, rendering their precise contribution to locomotion difficult to assess. We show in this Letter that a single passive filament hinged on the rear of a bluff body placed in a stream can generate a net lift force without increasing the mean drag force on the body. This is a consequence of spontaneous symmetry breaking in the filament's flapping dynamics. The phenomenon is related to a resonance between the frequency associated with the von Kármán vortex street developing behind the bluff body and the natural frequency of the free bending vibrations of the filament.
We analyze the effects of different types and positions of actuators and sensors on controllers' performance and robustness in the linearized 2D Blasius boundary layer. The investigation is carried out using direct numerical simulations (DNS). To facilitate controller design, we find reduced-order models from the DNS data using a system identification procedure called the Eigensystem Realization Algorithm. Due to the highly convective nature of the boundary layer and corresponding time delays, the relative position of the actuator and sensor has a strong influence on the closed-loop dynamics. We address this issue by considering two different configurations. When the sensor is upstream of the actuator, corresponding to disturbance-feedforward control, good performance is observed, as in previous work. However, feedforward control can be degraded by additional disturbances or uncertainties in the plant model, and we demonstrate this. We then examine feedback controllers in which the sensor is a short distance downstream of the actuator. Sensors farther downstream of the actuator cause inherent time delays that limit achievable performance. The performance of the resulting feedback controllers depends strongly on the form of actuation introduced, the quantities sensed, and the observability of the structures deformed by the controller's action. These aspects are addressed by varying the spatial distribution of actuator and sensor. We find an actuator-sensor pair that is well-suited for feedback control, and demonstrate that it has good performance and robustness, even in the presence of unmodeled disturbances.
Recent progress in understanding subcritical transition to turbulence is based on the concept of the edge, the manifold separating the basins of attraction of the laminar and the turbulent state. Originally developed in numerical studies of parallel shear flows with a linearly stable base flow, this concept is adapted here to the case of a spatially developing Blasius boundary layer. Longer time horizons fundamentally change the nature of the problem due to the loss of stability of the base flow due to Tollmien-Schlichting (TS) waves. We demonstrate, using a moving box technique, that efficient long-time tracking of edge trajectories is possible for the parameter range relevant to bypass transition, even if the asymptotic state itself remains out of reach. The flow along the edge trajectory features streak switching observed for the first time in the Blasius boundary layer. At long enough times, TS waves co-exist with the coherent structure characteristic of edge trajectories. In this situation we suggest a reinterpretation of the edge as a manifold dividing the state space between the two main types of boundary layer transition, i.e. bypass transition and classical transition.
Adverse pressure-gradient (APG) turbulent boundary layers (TBL) are studied by performing well-resolved large-eddy simulations. The pressure gradient is imposed by defining the free-stream velocity distribution with the description of a power law. Different inflow conditions, box sizes and upper boundary conditions are tested in order to determine the final set-up. The statistics of turbulent boundary layers with two different power-law coefficients and thus magnitudes of adverse pressure gradients are then compared to zero pressure-gradient (ZPG) data. The effect of the APG on TBLs is manifested in the mean flow through a much more prominent wake region and in the Reynolds stresses through the existence of an outer peak. The pre-multiplied energy budgets show, that more energy is transported from the near-wall region to farther away from the wall.
An atmospheric boundary layer (ABL) is generally a very high Reynolds number boundary layer over a fully rough surface that is influenced by different external forces. Numerical simulations of ABLs are typically demanding, particularly due to the high Reynolds numbers. Large eddy simulation (LES) where the grid filtered Navier--Stokes equations are solved together with a turbulence model for the subgrid-scale motions is the most accurate and widely used technique to date for ABLs. However, high Reynolds numbers, filtered equations and rough surfaces do not support the simple no-slip boundary conditions together with a feasible grid resolution. A paramount part for the performance of an ABL LES simulation therefore lies in the quality of approximate wall boundary conditions, so called wall models.
The vast majority of LES codes used for ABL simulations rely on spatial discretization methods with low order finite difference approximations for the derivatives in the inhomogeneous wall normal direction. Furthermore, the wall boundary conditions are typically chosen in a mesh-dependent, non-local way, relying on the finite differences formulation.
In this thesis we focus on solving the ABL LES equations with a fully (pseudo) spectral Fourier--Chebyshev code. We present how wall boundary conditions can be formulated through Robin boundary conditions and how to implement these in the normal-velocity normal-vorticity formulation that we solve. A new idea of specifying boundary conditions directly in Fourier space where also the turbulence intensity statistics can be controlled is presented and verified. The present results show that the Robin-type formulation is effective at least in near-equilibrium boundary layers.
The code and boundary conditions were tested in both low and high Reynolds number (open and full) channel flows of neutral and stable stratification. Results were validated with both low to moderate Reynolds number DNS statistics as well as with the logarithmic law. Our results indicate great potential for both the the new boundary condition formulation and the specific code implementation. Further analysis of more complex flow situations will show whether the Robin-type formulation will give similarly good results.
The experimental configuration in [M. Asai, M. Minagawa, M. Nishioka, The instability and breakdown of a near-wall low-speed streak, J. Fluid Mech. 455 (2002) 289-314] is numerically reproduced in order to examine the instability of a single low-speed streak in a laminar boundary layer and to investigate the resulting generation of coherent structures. Such a configuration is chosen since the experimental data show that the two instability modes, varicose and sinuous, are of comparable strength. The instability characteristics are retrieved from the simulation of the flow impulse response. The varicose instability is associated to higher frequencies and lower group velocities than those of the sinuous modes. The latter are less affected by the diffusion of the streak mean shear and are amplified for a longer streamwise distance. Analysis of the perturbation kinetic energy production reveals that both the varicose and the sinuous instability are driven by the work of the Reynolds stress against the wall-normal shear of the streak. The base flow considered here therefore presents an exception to the common knowledge, supported by several previous studies, that the sinuous instability is associated to the streak spanwise shear. The vortical structures at the late stage of the varicose breakdown are identified from the numerical data. By comparing them with those pertaining to other transition scenarios, it is confirmed that streaks and streamwise vortices are universal features of boundary layer transition.
The objective of this paper is to show that the interaction of streamwise velocity streaks of finite length can lead to turbulent breakdown in the flat-plate boundary layer flow. The work is motivated by previous numerical and experimental studies of transitional flows where the high-frequency oscillations leading to turbulence are seen to form in the region of strongest shear induced by streaks in relative motion. Therefore, a model for the interaction of steady and unsteady (i.e., slowly moving in the spanwise direction) spanwise periodic streaks is proposed. The interaction of two subsequent streaks is investigated for varying collision parameters. In particular, the relative spanwise position and angle are considered. The results show that the interaction is able to produce both a symmetric and asymmetric breakdown without the need for additional random noise from the main stream. Velocity structures characteristic of both scenarios are analyzed. Hairpin and A vortices are found in the case of symmetric collision between a low-speed region and an incoming high-speed streak, when a region of strong wall-normal shear is induced. Alternatively, when the incoming high-momentum fluid is misaligned with the low-speed streak in front, single quasi-streamwise vortices are identified. Despite the different symmetry at the breakdown, the detrimental interaction involves for both cases the tail of a low-speed region and the head of a high-speed streak. Further, the breakdown appears in both scenarios as an instability of three-dimensional shear layers formed between the two streaks. The streak interaction scenario is suggested to be of relevance for turbulence production in wall-bounded flows.
Non-modal analysis determines the potential for energy amplification in stable flows. The latter is quantified in the frequency domain by the singular values of the resolvent operator. The present work extends previous analysis on the effect of base-flow modifications on flow stability by considering the sensitivity of the flow non-modal behaviour. Using a variational technique, we derive an analytical expression for the gradient of a singular value with respect to base-flow modifications and show how it depends on the singular vectors of the resolvent operator, also denoted the optimal forcing and optimal response of the flow. As an application, we examine zero-pressure-gradient boundary layers where the different instability mechanisms of wall-bounded shear flows are all at work. The effect of the component-type non-normality of the linearized Navier-Stokes operator, which concentrates the optimal forcing and response on different components, is first studied in the case of a parallel boundary layer. The effect of the convective-type non-normality of the linearized Navier-Stokes operator, which separates the spatial support of the structures of the optimal forcing and response, is studied in the case of a spatially evolving boundary layer. The results clearly indicate that base-flow modifications have a strong impact on the Tollmien-Schlichting (TS) instability mechanism whereas the amplification of streamwise streaks is a very robust process. This is explained by simply examining the expression for the gradient of the resolvent norm. It is shown that the sensitive region of the lift-up (LU) instability spreads out all over the flat plate and even upstream of it, whereas it is reduced to the region between branch I and branch II for the TS waves.
Large-scale instabilities occurring in the presence of small-scale turbulent fluctuations are frequently observed in geophysical or astrophysical contexts but are difficult to reproduce in the laboratory. Using extensive numerical simulations, we report here on intense recurrent bursts of turbulence in plane Poiseuille flow rotating about a spanwise axis. A simple model based on the linear instability of the mean flow can predict the structure and time scale of the nearly periodic and self-sustained burst cycles. Poiseuille flow is suggested as a prototype for future studies of low-dimensional dynamics embedded in strongly turbulent environments.
This thesis considers the stability and transition of incompressible boundary layers. In particular, the Falkner–Skan–Cooke boundary layer subject to a cylindrical surface roughness, and the Blasius boundary layer with applied localized suction are investigated. These flows are of great importance within the aviation industry, feature complex transition scenarios, and are strongly three-dimensional in nature. Consequently, no assumptions regarding homogeneity in any of the spatial directions are possible, and the stability of the flow is governed by an extensive three-dimensional eigenvalue problem.
The stability of these flows is addressed by high-order direct numerical simulations using the spectral element method, in combination with a Krylov subspace projection method. Such techniques target the long-term behavior of the flow and can provide lower limits beyond which transition is unavoidable. The origin of the instabilities, as well as the mechanisms leading to transition in the aforementioned cases are studied and the findings are reported.
Additionally, a novel method for computing the optimal forcing of a dynamical system is developed. This type of analysis provides valuable information about the frequencies and structures that cause the largest energy amplification in the system. The method is based on the inverse power method, and is discussed in the context of the one-dimensional Ginzburg–Landau equation and a two-dimensional flow case governed by the Navier–Stokes equations.
This thesis considers the hydrodynamic instability and optimal forcing of a number of incompressible flow cases. In the first part, the instabilities of three problems that are of great interest in energy and aerospace applications are studied, namely a Blasius boundary layer subject to localized wall-suction, a Falkner–Skan–Cooke boundary layer with a localized surface roughness, and a pair of helical vortices. The two boundary layer flows are studied through spectral element simulations and eigenvalue computations, which enable their long-term behavior as well as the mechanisms causing transition to be determined. The emergence of transition in these cases is found to originate from a linear flow instability, but whereas the onset of this instability in the Blasius flow can be associated with a localized region in the vicinity of the suction orifice, the instability in the Falkner–Skan–Cooke flow involves the entire flow field. Due to this difference, the results of the eigenvalue analysis in the former case are found to be robust with respect to numerical parameters and domain size, whereas the results in the latter case exhibit an extreme sensitivity that prevents domain independent critical parameters from being determined. The instability of the two helices is primarily addressed through experiments and analytic theory. It is shown that the well known pairing instability of neighboring vortex filaments is responsible for transition, and careful measurements enable growth rates of the instabilities to be obtained that are in close agreement with theoretical predictions. Using the experimental baseflow data, a successful attempt is subsequently also made to reproduce this experiment numerically.
In the second part of the thesis, a novel method for computing the optimal forcing of a dynamical system is developed. The method is based on an application of the inverse power method preconditioned by the Laplace preconditioner to the direct and adjoint resolvent operators. The method is analyzed for the Ginzburg–Landau equation and afterwards the Navier–Stokes equations, where it is implemented in the spectral element method and validated on the two-dimensional lid-driven cavity flow and the flow around a cylinder.
In this work the problem of premature transition in boundary layers due to localized suction is revisited. A thorough study involving nonlinear direct numerical simulations, a three-dimensional linear stability analysis, a sensitivity study and a Koopman analysis is presented. The ensemble of these different techniques enables the origins of oversuction to be studied in great detail and provides new insight into the transition process of the flow. The configuration considered consists of an infinite row of widely separated suction pipes that are mounted to the plate at right angles. For the parameter range investigated, the flow inside the pipe is seen to bifurcate at a lower suction ratio than the boundary layer and thus act as an oscillator that forces the external flow over the plate. At low levels of suction, this forcing is not enough to cause transition in the boundary layer, but as the suction level is increased beyond criticality, modes originating from the pipe and extending into the boundary layer are seen to destabilize as well. These modes enable the perturbations forced in the pipe to also amplify in the boundary layer, which leads to a rapid breakdown to turbulence in the wake of the suction hole.
The stability of the Blasius boundary layer subject to localized suction is revisited. Using tools of global stability analysis, the leading direct and adjoint eigenpairs are determined, and novel insight into the sensitivity and receptivity of the flow is obtained. The problem is addressed through high-order spectral element simulations, which enables the inclusion of a suction pipe into the domain. Due to this, a detailed investigation of the connection between the pipe flow and the boundary layer flow is possible. For all cases investigated, the former always turns out to transition for a lower Reynolds number and suction rate than the latter, and the transition scenario is found to be due to a global instability originating inside a separation bubble at the pipe inlet. Identification of such regions, provides information that is valuable in further development of algorithms for laminar flow control.
The need to numerically represent a free vortex system arises frequently in fundamental and applied research. Many possible techniques for realizing this vortex system exist but most tend to prioritize accuracy either inside or outside of the vortex core, which therefore makes them unsuitable to for a stability analysis considering the entire flow field. In this article, a simple method is presented that is shown to yield an accurate representation of the flow inside and outside of the vortex core. The method is readily implemented in any incompressible Navier–Stokes solver using primitive variables and Cartesian coordinates. It can potentially be used to model a wide range of vortices but is here applied to reproduce a recent experiment by Quaranta et al. (2017) considering two helices. A three-dimensional stability analysis is performed and yields an eigenvalue spectrum that features both long- and short-wave instabilities.
A three-dimensional global stability analysis using high-order direct numerical simulations is performed to investigate the effect of surface roughness with Reynolds number (based on roughness height) Rek above and below the critical value for transition, on the eigenmodes of a Falkner-Skan-Cooke boundary layer. The surface roughness is introduced with the immersed boundary method and the eigenvalues and eigenfunctions are solved using an iterative time-stepper method. The study reveals a global instability for the case with higher Reynolds number that causes the flow in the non-linear simulations to break down to turbulence shortly downstream of the roughness. Examination of the unstable linear global modes show that these are the same modes that are observed in experiments immediately before breakdown due to secondary instability, which emphasizes the importance of these modes in transition.
A global stability analysis of a Falkner-Skan-Cooke boundary layer with distributed three-dimensional surface roughness is per- formed using high-order direct numerical simulations. Computations have been performed for different sizes of the roughness elements, and a time-stepping method has been used to find the instability modes. The study shows that a critical roughness height beyond which a global instability is excited does exist. Furthermore, the origins of this instability is examined by means of an energy analysis, which reveals the production and dissipation terms responsible for the instability, as well as the region in space where the instability originates.
A three-dimensional linear global stability analysis of a Falkner–Skan–Cooke boundary layer with distributed three-dimensional surface roughness is performed. The Falkner–Skan–Cooke boundary layer models the flow over swept airplane wings, and investigation of the critical roughness size for which a global instability emerges is thus of great importance within aeronautical applications. The study considers high-order direct numerical simulations and shows that such a critical roughness height exists for the Falkner–Skan–Cooke boundary layer. The roughness Reynolds number and roughness element aspect ratio for which this happens is comparable to the transition data reported in the literature for two-dimensional boundary layers. This demonstrates the importance of the local flow conditions in the vicinity of the roughness for triggering a global instability, although the resulting breakdown scenario is completely different from that of two-dimensional boundary layers. This breakdown scenario is studied in detail, and a global energy analysis is used to reveal the structures and mechanisms responsible for production and dissipation of perturbation energy.
With the motivation of determining the critical roughness size, a global stability and sensitivity analysis of a three-dimensional Falkner–Skan–Cooke (FSC) boundary layer with a cylindrical surface roughness is performed. The roughness size is chosen such that breakdown to turbulence is initiated by a global version of traditional secondary instabilities of the crossflow (CF) vortices, instead of an immediate flow tripping at the roughness. The resulting global eigenvalue spectra of the systems are found to be very sensitive to numerical parameters and domain size. This sensitivity to numerical parameters is quantified using the "-pseudospectrum, and the dependency on the domain is analysed through an impulse response and an energy budget. It is shown that the growth rates increase with domain size, which originates from the inclusion of stronger CF vortices in the baseflow. This is reflected in a change in the rate of advective energy transport by the baseflow. It is concluded that the onset of global instability in a FSC boundary layer as the roughness height is increased does not correspond to an immediate flow tripping behind the roughness, but occurs for lower roughness heights if su ciently long domains are considered. However, the great sensitivity results in an inability to accurately pinpoint the exact parameter values for the bifurcation, and the large spatial growth of the disturbances in the long domains eventually becomes larger than what can be resolved using finite precision arithmetics.
For problems governed by a non-normal operator, the leading eigenvalue of the operator is of limited interest and a more relevant measure of the stability is obtained by considering the harmonic forcing causing the largest system response. Various methods for determining this so-called optimal forcing exist, but they all suffer from great computational expense and are hence not practical for large-scale problems. In the present paper a new method is presented, which is applicable to problems of arbitrary size. The method does not rely on timestepping, but on the solution of linear systems, in which the inverse Laplacian acts as a preconditioner. By formulating the search for the optimal forcing as an eigenvalue problem based on the resolvent operator, repeated system solves amount to power iterations, in which the dominant eigenvalue is seen to correspond to the energy amplification in a system for a given frequency, and the eigenfunction to the corresponding forcing function. Implementation of the method requires only minor modifications of an existing timestepping code, and is applicable to any partial differential equation containing the Laplacian, such as the Navier-Stokes equations. We discuss the method, first, in the context of the linear Ginzburg-Landau equation and then, the two-dimensional lid-driven cavity flow governed by the Navier-Stokes equations. Most importantly, we demonstrate that for the lid-driven cavity, the optimal forcing can be computed using a factor of up to 500 times fewer operator evaluations than the standard method based on exponential timestepping.
For problems governed by a non-normal operator, the leading eigenvalue of the operator is of limited interest and a more relevant measure of the stability is obtained by considering the harmonic forcing causing the largest system response. Various methods for determining this so-called optimal forcing exist, but they all suffer from great computational expense and are hence not practical for large-scale problems. In the present paper a new method is presented, which is applicable to problems of arbitrary size. The method does not rely on timestepping, but on the solution of linear systems, in which the inverse Laplacian acts as a preconditioner. By formulating the problem of finding the optimal forcing as an eigenvalue problem based on the resolvent operator, repeated system solves amount to power iterations, in which the dominant eigenvalue is seen to correspond to the energy amplification in a system for a given frequency, and the eigenfunction to the optimal forcing function. Implementation of the method requires only minor modifications of an existing time-stepping code, and is applicable to any partial differential equation containing the Laplacian, such as the Navier-Stokes equations. We discuss it in the context of the linear Ginzburg-Landau equation.
This work is concerned with a detailed investigation of the steady (laminar), incompressible flow inside bent pipes. In particular, a toroidal pipe is considered in an effort to isolate the effect of the curvature, δ, on the flow features, and to compare the present results to available correlations in the literature. More than 110 000 numerical solutions are computed, without any approximation, spanning the entire curvature range, 0 ≤ δ ≤ 1, and for bulk Reynolds numbers Re up to 7 000, where the flow is known to be unsteady. Results show that the Dean number De provides a meaningful non-dimensional group only below very strict limits on the curvature and the Dean number itself. For δ>10−6 and De > 10, in fact, not a single flow feature is found to scale well with the Dean number. These considerations are also valid for quantities, such as the Fanning friction factor, that were previously considered Dean-number dependent only. The flow is therefore studied as a function of two equally important, independent parameters: the curvature of the pipe and the Reynolds number. The analysis shows that by increasing the curvature the flow is fundamentally changed. Moderate to high curvatures are not only quantitatively, but also qualitatively different from low δ cases. A complete description of some of the most relevant flow quantities is provided. Most notably the friction factor f for laminar flow in curved pipes by Ito [J. Basic Eng. 81:123–134 (1959)] is reproduced, the influence of the curvature on f is quantified and the scaling is discussed. A complete database including all the computed solutions is available at www.flow.kth.se.
A complete original study of the linear temporal instability analysis of two viscous and immiscible fluids enclosed in a rigid cylinder rotating about its axis and separated by a cylindrical interface is performed for the case of higher density fluid located in the annulus. The results of the present contribution fill the lack of an overall assessment of the system behavior due to the increase of both the analytical difficulties and the number of the governing parameters when the several physical effects are all included. The analysis is carried out numerically by discretizing the equations of the evolution of disturbances separately in the two phases formulated in a rotating reference frame. Normal mode analysis leads to a generalized eigenvalue problem which is solved by means of a Chebyshev collocation spectral method. The investigation of the preferred modes of instability is carried out over wide ranges of the parameters space. The behavior of the system is physically discussed and is compared to inviscid asymptotic limits and to viscous approximate solutions of the previous literature.
The optimal growth of perturbations to transiently growing streaks is studied in Poiseuille flow. Basic flows are generated by direct numerical simulation giving primary optimal spanwise periodic vortices of finite amplitude as the initial condition. They evolve into finite amplitude primary transiently growing streaks. Linear secondary optimal energy growth supported by these primary flows are computed using an adjoint technique which takes into full account the unsteadiness of the basic flows. Qualitative differences between primary and secondary optimal growths are found only when the primary streaks are locally unstable. For locally stable primary streaks, the secondary optimal growth has the same scalings with Reynolds number R as the primary optimal growth and the maximum growth is attained by streamwise uniform vortices, suggesting that the primary and secondary optimal growth are based on the same physical mechanisms. When the primary streaks are locally unstable the secondary optimal growth of unstable wavenumbers scale differently with R and the maximum growth is attained for streamwise nonuniform sinuous perturbations, indicating the prevalence of the inflectional instability mechanism.
In this thesis direct numerical simulation is used to investigate the possibilityto delay the transition from laminar to turbulent in boundary layer flows.Furthermore, modal analysis is used to reveal the coherent structures in highdimensional dynamical systems arising in the flow problems.Among different transition scenarios, the classical transition scenario isanalysed. In this scenario, the laminar-turbulent transition occurs when Tollmien-Schlichting waves are triggered inside the boundary layer and grow exponentiallyas they move downstream in the domain. The aim is to attenuate the amplitudeof these waves using active control strategy based on a row of spatiallylocalised sensors and actuators distributed near the wall inside the boundarylayer. To avoid the high dimensional system arises from discretisation of theNavier Stokes equation, a reduced order model (ROM) based on EigensystemRealisation Algorithm (ERA) is obtained and a linear controller is designed.A plasma actuator is modelled and implemented as an external forcing on theflow. To account for the limitation of the plasma actuators and to further reducethe complexity of the controller several control strategies are examinedand compared. The outcomes reveal successful performance in mitigating theenergy of the disturbances inside the boundary layer.To extract coherent features of the wind turbine wakes, modal decompositiontechnique is employed where a large scale dynamical system is reduced toa fewer number of degrees of freedom. Two decomposition techniques are employed:proper orthogonal decomposition and dynamic mode decomposition.In the former procedure, the flow is decomposed into a set of uncorrelated structureswhich are rank according to their energy. In the latter, the eigenvaluesand eigenvectors of the underlying approximate linear operator is computedwhere each mode is associated with a specific frequency and growth rate. Theresults revealed the structures which are dynamically significant to the onsetof instability in the wind turbine wakes.
In this thesis direct numerical simulation is used to investigate the possibility to delay the transition from a laminar to a turbulent flow in boundary layer flows. Furthermore, modal analysis techniques are used to identify the coherent structures in wind turbine wake.
Among different transition scenarios, the classical transition scenario is considered where the laminar-turbulent transition occursdue to Tollmien-Schlichting (TS) waves. These waves are convectively unstable and when triggered inside the boundary layer, they grow exponentially inamplitude as they are advected downstream of the domain. The aim is to suppressthese waves using flow control strategies based on a row of spatially localised sensors and actuators distributed near the wall inside the boundary layer. To avoid the high dimensionality arising from discretisation of the Navier–Stokes equations, a reduced order model (ROM) based on the Eigensystem Realisation Algorithm(ERA) is obtained and based on that a linear controller is designed. To manip-ulate the flow, a plasma actuator is modelled and implementedas an externalforcing. To account for the restrictions of the plasma actuators, several strategies are proposed and tested within the LQG framework. We studied also the design of a faster controller based on decentralised approach and compared the performance to a more expensive centralised controller. The outcomes revea lsuccessful performance in mitigating the energy of the disturbances inside the boundary layer and suppressing the TS waves.
To extract coherent features of the wind turbine wakes, modal decomposition techniques are employed. In this method a large dynamical system is reduced to a fewer number of degrees of freedom. Two decomposition techniques are employed, namely proper orthogonal decomposition and dynamic mode decomposition. In the former, the flow is decomposed into a set of orthogonal structures which are ranked according to their energy contents in a hierarchical manner. In the latter, the eigenvalues and eigenvectors of the underlying approximate linear operator of the system is evaluated. In particular each mode is associated with a specific frequency and growth rate. The results revealed the coherent structures which are dynamically significant for the onset of instability in the wind turbine wake
We use linear control theory to construct an output feedback controller for the attenuation of small-amplitude three-dimensional Tollmien-Schlichting (TS) wavepackets in a flat-plate boundary layer. A three-dimensional viscous, incompressible flow developing on a zero-pressure gradient boundary layer in a low Reynolds number environment is analyzed using direct numerical simulations. In this configuration, we distribute evenly in the spanwise direction up to 72 localised objects near the wall (18 disturbances sources, 18 actuators, 18 estimation sensors and 18 objective sensors). In a fully three-dimensional configuration, the interconnection between inputs and outputs becomes quickly unfeasible when the number of actuators and sensors increases in the spanwise direction. The objective of this work is to understand how an efficient controller may be designed by connecting only a subset of the actuators to sensors, thereby reducing the complexity of the controller, without comprising the efficiency. If n and m are the number of sensor-actuator pairs for the whole system and for a single control unit, respectively, then in a decentralised strategy, the number of interconnections deceases mn compared to a centralized strategy, which has n (2) interconnections. We find that using a semi-decentralized approach - where small control units consisting of 3 estimation sensors connected to 3 actuators are replicated 6 times along the spanwise direction - results only in a 11 % reduction of control performance. We explain how "wide" in the spanwise direction a control unit should be for a satisfactory control performance. Moreover, the control unit should be designed to account for the perturbations that are coming from the lateral sides (crosstalk) of the estimation sensors. We have also found that the influence of crosstalk is not as essential as the spreading effect.
An active control strategy is implemented to attenuate the amplitude of the Tollmien-Schlichting (TS) waves inside the boundary layer of an airfoil. The dynamics of the system are modelled by the linearised Navier-Stokes equations. The impulse response to an initial disturbance, initially located outside of the boundary layer and in front of the airfoil is considered. The perturbation evolves and penetrates inside the boundary layer and triggers the TS waves. Different control strategies including the linear quadratic Gaussian (LQG) and model predictive control (MPC) are designed based on a reduced order model where the sensors and actuators are localised near the wall. An output projection is used to identify the unstable disturbances; the objective function of the controller is selected as a set of proper orthogonal decomposition (POD) modes; to isolate the dynamics of the TS waves, the modes with high energy contents in the TS wave frequency band are considered as the objective of the controller. A plasma actuator is modelled and implemented as an external forcing on the flow. To account for the limitations of the plasma actuator several strategies are examined and the results are compared with a classical LQG controller. The outcomes reveal successful performance in mitigating the amplitude of the wavepacket developing inside the boundary layer.
The evolution and control of a two dimensional (2D) wavepacket developing on a flat plate with a leading edge is investigated by means of direct numerical simulation (DNS).
The aim is to identify and suppress the wavepackets generated by freestream perturbations. A sensor is placed close to the wall in order to detect the upcoming wavepacket, while an actuator is placed further downstream to control it. A plasma actuator is modelled as an external forcing on the flow using a model based and validated on experimental investigations. A Linear Quadratic Gaussian (LQG) controller is designed and an output projection is used to build the objective function. Moreover, by appropriate selection of the Proper Orthogonal Decomposition (POD) modes, we identify the disturbances to be damped. A reduced-order model of the input-output system is constructed by using system identification via the Eigensystem Realization Algorithm (ERA) algorithm.
A limitation of the plasma actuators is the uni-directional forcing of the generated wall jet, which is predetermined by the electrodes location. In this paper, we address this limitation by proposing and comparing two different solutions: i) by introducing an offset in the control signal such that the resulting total forcing is oriented along one direction; ii) by using two plasma actuators acting in opposite directions. The results are compared with the ideal case where constraints are not accounted for the control design. We show that the resulting controllers based on plasma actuators can successfully attenuate the amplitude of the wavepacket developing inside the boundary layer.
The evolution and control of a two-dimensional wave packet developing on a flat plate with a leading edge is investigated by means of direct numerical simulation. The aim is to identify and suppress the wave packets generated by freestream perturbations. A sensor is placed close to the wall to detect the upcoming wave packet, while an actuator is placed further downstream to control it. A plasma actuator is modeled as an external forcing on the flow using a model based and validated on experimental investigations. A linear quadratic Gaussian controller is designed, and an output projection is used to build the objective function. Moreover, by appropriate selection of the proper orthogonal decomposition modes, we identify the disturbances to be damped. A reduced-order model of the input-output system is constructed by using system identification via the eigensystem realization algorithm. A limitation of the plasma actuators is the unidirectional forcing of the generated wall jet, which is predetermined by the electrodes' location. In this paper, we address this limitation by proposing and comparing two different solutions: 1) introducing an offset in the control signal such that the resulting total forcing is oriented along one direction, and 2) using two plasma actuators acting in opposite directions. The results are compared with the ideal case where constraints are not accounted for the control design. We show that the resulting controllers based on plasma actuators can successfully attenuate the amplitude of the wave packet developing inside the boundary layer.
Nonmodal instability analysis is carried out for a 2:1 elliptic cone with base flow conditions selected for a Ma=7 and two different ight altitudes, namely 33km and 21km with unit Reynolds number Re′ = 1.89 x 106 m-1 and Re′ = 1.015 x 107 m-1, respectively. The aim is to analyze the effects of transiently growing optimal disturbances and their possible relation to instability mechanisms that have been confirmed to exist in previous modal crossow. Local linear stability results obtained at several streamwise locations on the cone surface indicate that transient growth in the crossow region may be correlated to streamwise oriented structures having spanwise spacing of the same order of magnitude as which have long been known to exist in this flow.
Subcritical transition to turbulence requires finite-amplitude perturbations. Using a nonlinear optimisation technique in a periodic computational domain, we identify the perturbations of plane Couette flow transitioning with least initial kinetic energy for Re <= 3000. We suggest a new scaling law E-c = O(Re-2.7) for the energy threshold vs. the Reynolds number, in quantitative agreement with experimental estimates for pipe flow. The route to turbulence associated with such spatially localised perturbations is analysed in detail for Re = 1500. Several known mechanisms are found to occur one after the other: Orr mechanism, oblique wave interaction, lift-up, streak bending, streak breakdown, and spanwise spreading. The phenomenon of streak breakdown is analysed in terms of leading finite-time Lyapunov exponents of the associated edge trajectory.
We present for the first time a complete bifurcation diagram of plane Couette flow based on direct numerical simulation of the full Navier-Stokes equations. The use of an unusually large computational domain (800h x 2h x 356h) is crucial for the determination of transition thresholds, because it allows to reproduce spatio-temporal intermittency structures such as transient spots, turbulent bands, and laminar holes. The threshold in Re (based on the half-gap) is found to he Re-c = 324 +/- 1 in very good agreement with available experimental data. This work points out that, at the onset of transition in Re, fragmented oblique patterns always emerge from the interaction of growing neighbouring spots. An analogy with thermodynamical phase transition seems relevant to describe the whole transition process.
Upstream flow deformation (UFD) has been shown to be an effective technique to delay roughness induced laminar-turbulent transition at low free-stream turbulence level in three-dimensional boundary-layer flows. Beneficial steady crossflow vortex (CFV) control modes are excited and the resulting nonlinear CFVs induce a useful mean-flow distortion. We recently showed by direct numerical simulations that plasma actuators, modeled by localized steady volume forcing, can be employed to excite the UFD control modes. In the current work we investigate the same actuator set-ups to control transition caused by traveling CFVs that are excited by single unsteady vortical free-stream disturbances (FSDs) impinging on the boundary layer. FSDs of various wavenumbers and frequencies are imposed either upstream or downstream, or at the position of the actuators to also scrutinize if the volume forcing has a direct unfavorable effect on the receptivity to the FSDs that adds to the stabilization by the UFD. For all investigated cases we show that a significant transition delay is achieved.