The control of flow separation by periodic excitation

Abstract

This paper presents a review of the control of flow separation from solid surfaces by periodic excitation. The emphasis is placed on experimentation relating to hydrodynamic excitation, although acoustic methods as well as traditional boundary layer control, such as steady blowing and suction, are discussed in order to provide an appropriate historical context for recent developments. The review examines some aspects of the excited plane mixing-layer and shows how its development lays the foundation for a basic understanding of the problem. Flow attachment to, and separation from, a deflected flap is then shown to be a paradigm for isolating controlling parameters as well as understanding the basic mechanisms involved. Particular attention is paid to separation control on airfoils by considering controlling parameters such as optimum reduced frequencies and excitation levels, performance enhancement, efficiency, reduction of post-stall unsteadiness, compressibility and other important features. Additional topics covered include excitation of separation bubbles, control and exploitation of diffuser flows, three-dimensional effects, the influence of longitudinal curvature and possible applications to unmanned air vehicles. The review closes with some recent developments in the control and understanding of incompressible dynamic stall, specifically illustrating the control of dynamic stall on oscillating airfoils and identifying the crucial time-scale disparity between dynamic stall and periodic excitation.

Introduction

Flow separation is generally accepted to be the breakaway or detachment of fluid from a solid surface (e.g. [1], [2], [3], [4]). Whether caused by a severe adverse pressure gradient (e.g. [5], [6]), a geometrical aberration (e.g. [7], [8]) or by any other means, separation is generally accompanied by significant thickening of the rotational flow region adjacent to the surface with a marked increase in the velocity component that is normal to the surface. Thus our traditional means for analysis, namely the boundary layer equations, are summarily invalidated. Separation is almost always associated with losses of some kind, including loss of lift, drag increase, pressure recovery losses, etc. Consequently, engineers have been preoccupied, for almost a century, with altering its location or avoiding it entirely. The multitude and variety of hydro and aerodynamic vehicles and devices, taken for granted today, bear testimony to the tremendous advances that have been made in the development of means that avoid or modify separation.

Notwithstanding these advances, the development of commercial and military vehicle aerodynamics has been severely stunted in recent decades. The underlying problem is a design philosophy which is steeped in a century old tradition: the steady-flow assumption. Conventional wisdom tells us that so-called steady two-dimensional separation is well understood, with a satisfactory analytical foundation (e.g. [4]). Experimental data reveals, however, that separation is only steady in a time-averaged sense, while the rich time-dependent coherence is acknowledged, but it is certainly not exploited. For highly turbulent flows which are on the verge of separation (e.g. [5], [9]), or known to have an underlying deterministic structure (e.g. turbo-machinery), the Reynolds-averaged equations have limited the designer to basic assumptions which, by their very definition, negate the phenomenon of time dependence.

Apart from the need to delay transition and control mixing, the familiar shapes of products which move through fluids or determine fluid motion are a direct consequence of the need to avoid separation. For example, the maximum thickness or camber of wings, and the length and shape of diffusers and inlets, evolved from this constraint whose norms are still dictated by the steady-flow assumption. For the better part of this century, methods for aircraft control have been rooted in the limiting steady-flow assumptions. A classical example is wing warping initiated by the Wright brothers (see, e.g. [10]), which evolved into ailerons, flaps and slats prominently displayed on today's aircraft (e.g. observe the multitude of moving surfaces on a Boeing 747 prior to landing). They are effective, but they are also complex, costly, cumbersome and heavy. Consider a maneuverable aircraft with no large moving parts or control surfaces, or a single-rotor helicopter with no tail rotor or no heavy NOTAR-type cylinder and compressor. Consider a conical diffuser whose divergence angle exceeds 45°, or thrust vectored jet without vectoring nozzles. All of these are possible, by adding periodic motion to the flow rather than traditional boundary layer control (BLC) methods, discussed below, which involve injecting or removing large amounts of momentum and mass in a steady fashion. In fact, periodic addition of momentum can attain the same degree of control authority as is achieved by traditional BLC, with one important difference: the cost, in terms of momentum input, is less by anything from factors to orders of magnitude [11].

One year after the first powered flight, Prandtl [12] introduced the concept of the boundary layer and proposed a means (i.e. suction through a slot) to control its attachment to a solid surface. The next ten years saw little progress in boundary layer research [13], but the race for air superiority, spawned universal research and development efforts in which BLC studies were given high a priority. The depth and breadth of international research up to the early 1960s is apparent from the two volumes edited by Lachmann [14], [15], which provided exhaustive treatment of theoretical, experimental as well as applied BLC methods. The contributions indicated that steady blowing or suction from various locations on the wing surface, and on various configurations, could produce significant increases in lift as well as reductions in drag. In fact, a large number of experimental aircraft were built, which convincingly demonstrated the effectiveness of BLC. Some of these aircraft reached mass production (for example Lockheed's F-104, which remained in production from 1955 to 1983 and the MIG-21 of which thousands still fly in the former Eastern Bloc and in “Non-Aligned” countries), but BLC fell far short of the high expectations of the 1960s. There were two main reasons for this: firstly, the plumbing systems required for BLC introduced excessive technical complexity and weight, and secondly, the systems were not efficient, requiring auxiliary compressors or excessive compressor bleed in order to obtain meaningful lift enhancements (see, e.g. [16]). Some BLC systems, such as that used on the MIG-21, utilize engine combustion gases and rely on the thermal inertia of the plumbing to withstand the heat for the short duration of application prior to landing.

The seeds for flow control by excitation were sewn by Schubauer and Skramstad [17] who introduced periodic perturbations in a laminar boundary layer to trigger a known instability, i.e. to initiate Tollmien-Schlichting (TS) waves. This breakthrough technique became a major diagnostic aid for studying flow stability but it was also considered a tool for controlling laminar separation and transition (e.g. [18], [19]). Since periodic perturbations trigger a premature transition to turbulence and a turbulent flow is less susceptible to separation, flow separation could be delayed by initiating transition. Sound was initially used to demonstrate these ideas on airfoils at low Reynolds numbers (see [20]). Although laminar–turbulent transition could be effected, the manipulation of turbulent shear flows was traditionally considered to be unattainable because of the belief that turbulence is a random process, determined largely by local flow conditions. However, experiments such as those of Brown and Roshko [21] and Winant and Browand [22] in the mixing layer demonstrated that large coherent structures (LCSs) are primarily responsible for the transport of momentum across the flow domain. Moreover, the introduction of periodic mechanical excitation at the flow origin was shown to be an efficient and convenient method for the control of mixing (e.g. [23], [24], [25]). Excitation accelerates and regulates the generation of large coherent structures, particularly when the mean flow is unstable, thereby transferring high momentum fluid across the mixing layer.

The last decade has witnessed progress in flow control where the fundamental studies of the 1970s and 1980s, such as those in free shear layers (see [26], [27]), set the scene for effective, efficient and practical separation control. Specifically, it was shown that a turbulent mixing layer could be attached to a deflected surface [28], [29] and the ensuing attached flow separation could be delayed [30], [31] by periodic addition of momentum with or without the concomitant addition of mass flux. The same concept was also applied to airfoils (e.g. [32], [34]) and was further extended to other applications. This method was shown to be much more effective than the traditional steady blowing and at times attained a saving of two orders of magnitude in the momentum coefficient required to achieve a prescribed improvement in performance. The actuators required may thus be autonomous, small, light, energy efficient and decoupled from the main propulsive systems – thereby overcoming all of the deficiencies of traditional BLC.

To date, separation control (acoustic or hydrodynamic) has been demonstrated in a wide variety of relatively simple configurations, such as flow over backward-facing steps and ramps (e.g. [35], [36]), on sharp leading-edge wedges [37], on bluff bodies (e.g. [38], [39]), on various airfoils (e.g. [40], [41], [42]), delta wings [43], circular cylinders (see e.g. [44]) and behind two-dimensional fences (e.g. [45]). These studies laid the groundwork for an extremely wide variety of potential future applications to both fixed wing (e.g. [46]) as well as rotary wing (e.g. [47]) aircraft.

The emphasis of this paper is on experimental data which either isolate parameters governing separation control or elucidate underlying physical mechanisms. This paper does not cover other issues related to separation control, including theoretical (e.g. [48], [49]) or computational prediction methods (e.g. [50], [51], [52], [216], [217]), energy-efficient micro-elecromechanical systems (MEMS – e.g. [53]) for actuation, “smart” or adaptive structures (e.g. [54]), separation detection methods (e.g. [55]) and control theory/neural-nets (e.g. [56], [57]). This paper also does not cover passive methods of separation control, even those which may exploit flow instabilities (see e.g. [4]). As the title suggests, the content of this paper is limited to flows where separation is controlled by excitation, i.e. an inherent instability is exploited to achieve the desired results. Consequently, separation control by blowing and suction or actuation which does not clearly exploit flow instability (e.g. [58], [59]), or similar methods which create pseudo-surfaces (e.g. [47]), are not considered to be within the scope of this paper.

In writing this article, the authors have endeavored to incorporate enough background material for the non-expert, while still maintaining adequate scope and depth for practitioners of the art. The authors also wish to point out that this review is unavoidably biased in favor of their own work as well as that of their colleagues. As such, important studies pertaining to separation control may have been unintentionally omitted. One positive aspect of this, however, is that relevant data, hitherto languishing in Hebrew-language graduate theses, is now available to the general reader.