Contributions of Particle Image Velocimetry to Helicopter Aerodynamics: A Review

Abstract

The advancement of flow measurement techniques continues to extend experimental boundaries and thus significantly contributes to improving our understanding of both basic and applied aerodynamics. This is particularly apparent in the case of particle image velocimetry (PIV), where its application has furthered the existing knowledge in several areas of helicopter rotor aerodynamics. The complex nature of helicopter rotor flows presents unique challenges to experimentalists, including transonic flow, concentrated vortices and dynamic stall. To illustrate the impact of the technological advancements on the way helicopter aerodynamics is studied today, the development of PIV since the early nineties of the last century is reviewed and some recent PIV applications are described. Using examples of main rotor wakes, dynamic stall and flow control investigations, the capabilities of large–scale, time–resolved and volumetric PIV are summarized.

1. INTRODUCTION

The unsteadiness and complexity of the flow field around helicopter rotor blades pose significant challenges for present aerodynamic investigations, as illustrated in Fig. 1. The concentrated blade tip vortices in the main rotor wake (Fig. 1b), compressible flow on the advancing blades (Fig. 1a), and dynamic stall on the retreating blades (Fig. 1c) affect the helicopter performance and limit the flight envelope (Conlisk 2001). All these phenomena must be addressed to optimize the performance of helicopters with respect to hovering efficiency, cruise speed, range, loading capacity, and aeroacoustic emissions. The complexity of rotor flow requires experimental investigations to go hand in hand with numerical simulations. In spite of the progress that has recently been made in the development of numerical prediction capabilities for isolated rotor components, detailed experiments are required to substantiate and confirm the results of complex CFD studies and to expand the understanding of flow phenomena related to helicopter flows (Allan et al. 2009, Antoniadis et al. 2012).

Fig. 1:

Some specific problems in experimental helicopter aerodynamics illustrated on DLR’s EC135 test helicopter, after Conlisk (2001). a) transonic flow on the advancing side, b) tip vortex formation, and c) dynamic stall on the retreating side of the main rotor

Early rotor flow field investigations were performed using flow visualization techniques and intrusive, point–wise techniques such as hot–wire anemometry and pressure probes (see e.g. Harris 1973). Since the early 1970s, point-wise, non-intrusive techniques such as the laser Doppler velocimetry (LDV) have been developed and applied for rotor measurements (Scully & Sullivan 1972, Boutier et al. 1996). However, the necessity to make measurements over an entire plane at a given instant of time led to the development of particle image velocimetry (PIV). Since the late 1990’s (Raffel et al. 1998c) PIV has become the state-of-the-art measurement technique for rotor flow field measurements. By using instantaneous measurements over an entire plane, the acquisition of unsteady flow features became experimentally feasible. The first applications of PIV on helicopter wind tunnel testing were made in the early 1990s and PIV has since undergone a rapid development, as indicated by the increasing number of published PIV-based rotor studies (Fig. 2).

Fig. 2:

Number of relevant articles investigating rotor blade tip vortex structure (estimated)

2. PARTICLE IMAGE VELOCIMETRY

2.1. Basic PIV principle

The principles of the basic PIV technique have been described in detail by Raffel et al (1998c) and Adrian and Westerweel (2011). A brief overview with a focus on the specific aspects of rotor applications is given here. A standard PIV setup is depicted in Fig. 3, featuring a particle-seeded flow field that is illuminated by a light source and imaged by a single camera at two consecutive instants in time. Such a configuration is defined as 2D PIV and the velocity field V is computed from the imaged particle displacements d = (d_x, d_y) between the two measurement images and the time delay (laser pulse separation time) ∆t as:

$$V=\frac{d}{M\Delta t}$$

where M is the magnification factor of the camera setup. The particle displacements are determined by splitting the measurement image into several small interrogation windows and computing the cross-correlation between corresponding windows in both image frames. It is understood that the cross-correlation algorithms used for the PIV evaluation are of statistical nature and represent the mean velocity of the particles within each interrogation window. The size of the interrogation windows thus has to be as small as possible to resolve high gradients in the flow field. This size reduction is limited by a minimum number of particles that has to be present in the interrogation window in both PIV frames. According to Adrian (1991), the particle displacement should therefore not exceed ¼ of the light sheet thickness, which limits the interframing time ∆t and thus the accuracy of the measurement. Restrictions concerning the in-plane displacement can be circumvented by applying adaptive grid interrogation schemes, such as the multi-grid method (Raffel et al. 1998c and references therein). The adaptive multi-grid scheme is based on the evaluation of a large particle displacement via a cascade of interrogation windows with successively decreasing size, which are spatially moved before subsequent iteration. The distance moved is based on the estimated displacement from the previous iteration. Thus only the coarse particle displacements in the initial large interrogation window have to be restricted to ¼ of the window size and are used to restrict the search area for a refined interrogation window of smaller size. This evaluation scheme thus combines high measurement accuracy through a large particle displacement and high measurement resolution through the small final interrogation windows.

Fig. 3:

Basic 2D-PIV setup in a wind tunnel, after Raffel et al. (1998c)

2.2. Three-component and volumetric PIV techniques

In spite of all its advantages, 2D PIV contains some shortcomings that necessitate further developments on the basis of instrumentation. One of these disadvantages is the fact that the 2D PIV method is only capable of recording the projection of the velocity vector into the plane of the light sheet; the out-of-plane velocity component is lost while the in-plane components are affected by an unrecoverable error due to the perspective transformation. For highly three-dimensional flows this can lead to substantial measurement errors of the local velocity vector. This error increases as the distance to the principal axis of the imaging optics increases. The most straightforward approach capable of recovering the complete set of velocity components in a given plane is an additional PIV recording from a different viewing direction using a second camera. Such a setup is generally referred to as stereoscopic PIV. Reconstruction of the three-component velocity vector in effect relies on the perspective distortion of a displacement vector viewed from different directions. A further refinement of the stereoscopic PIV method relies on the measurement of three components of velocities in multiple parallel planes, often referred to as dual- or multi-plane PIV. Multi-plane measurements thus yield all nine velocity gradients, which are e.g. required for turbulence estimation in a flow field (Ramasamy et al. 2009). Such a measurement can be achieved using two independent stereoscopic PIV setups or a combination of stereoscopic and 2D PIV (assuming continuity in the flow).

More recent efforts have been directed toward the recovery of full volume data sets using multiple camera digital imaging along with tomographic reconstruction. The volume PIV technique, also known as tomographic PIV, simultaneously images a given volume with at least three high resolution cameras from different directions. The over-determined nature of the image data allows for a tomographic reconstruction of the particle positions in space, such that a pair of these volume images recorded in short succession can be used to obtain a three-dimensional distribution of three-component velocity data. In practice the technique often makes use of four cameras which significantly increases the data yield with respect to three cameras. The low illumination levels due to volume illumination require large camera lens apertures which again limits the depth of field, such that a rather thick light sheet provides the best results. Detailed descriptions of the 3D PIV techniques are described by Raffel et al. (2007) Adrian and Westerweel (2011).

2.3. Limitations of PIV

The optical camera setup can reach its physical limit for small and medium sized apertures, which is also referred to as diffraction limited imaging. For large apertures (small f-numbers) lens distortions became more dominant. As a consequence, particle image diameters are rarely below 10 µm corresponding to 1–2 pixels of modern CCD cameras, but to less than 1 pixel for high-framing rate CMOS cameras. The mean particle displacement between two interrogation windows I₁, I₂ is evaluated by detecting a peak in the corresponding correlation map C_1,2, as shown on the left side of Fig. 4. The uncertainty in determining the imaged particle displacements depends on the precision of this peak detection, which can reach 0.01 pixels by using subpixel correlation peak detection algorithms (Raffel et al. 2007). In order to achieve this subpixel accuracy, a minimum recorded particle image diameter of ≥ 2 pixel is required to avoid the so-called peak locking effect. Peak locking describes the tendency of a continuous distribution of particle displacements to be shifted towards integer pixel values, and is caused by the discrete size of the image sensor. Larger particles also evoke increased intensity of the refracted laser light and thus improved signal for the cross correlation evaluation. While the restrictions imposed by PIV necessitate large particles, smaller particles are ideal to follow the flow and exhibit less inertial effects. These conflicting requirements in terms of seed particles should be considered when measuring velocities in a flow domain with high vorticity and inhomogeneous seeding density.

Fig. 4:

Raw particle image intensities in a PIV recording of a young rotor blade tip vortex (Ψ = 6°) with a particle void of diameter ∆A_c within the core are displayed on the left. The right hand side shows the enlarged interrogation windows I₁ and I₂, their asymmetric intensity distribution and the resulting correlation C_1,2

Another significant particle-related limitation of PIV occurs for velocity measurements close to compressible tip vortex cores. In the center of the vortex core the PIV evaluation fails entirely due to an insufficient number of imaged seeding particles. Inhomogeneous particle distributions in partly affected interrogation windows yield weak noise correlations at zero displacement. Due to the averaging nature of the finite sampling (i.e. interrogation) window sizes, the biasing effect is smeared out over a larger area than the void region itself (Fig. 4).

Despite these artifacts of PIV evaluation due to seeding particles, the peak tangential velocities detected by PIV are sometimes higher than predicted by CFD, indicating that neither flow measurement techniques nor numerical methods are able yet to accurately determine the true peak velocities around young vortices.

New evaluation methods will be needed to overcome the shortcomings of tracer particle based measurement methods. Current ideas to handle core particle voids consider localized interpolation procedures using low–order vortex models, similar to vortex parameter extraction methods. These methods are, however, restricted to the core region and have to be completed by a further refinement of PIV evaluation procedures including particle density weighted correlation procedures, which might also contribute to improved measurement performance for very small and strong vortices which are present at young wake ages.

3. EARLY PIV EXPERIMENTS ON HELICOPTER ROTORS

Until the mid-1990s, aerodynamic PIV measurements were rare and most commonly carried out by optical autocorrelation analysis of double-exposed photographic film. This technique was first applied to investigate helicopter aerodynamics with a study of the unsteady flow field around airfoils, simulating the unsteady transonic flow on the advancing blade (Raffel et al. 1993) and the dynamic stall on the retreating blade (Raffel and Kompenhans 1993, Geissler et al. 1993, Crisler et. al. 1994, Raffel et al. 1995a). First BVI investigations were performed photographically shortly thereafter (Horner et al. 1995,1996, Raffel et al. 1995a,b). The advent of full–frame progressive scan CCD sensors in 1996 enabled the realization of digital cross-correlation PIV. This technique is based on the recording of particle images in two separate frames with short inter-framing times and a computer-aided cross-correlation analysis of the image pairs. Directional ambiguity in the measurements, which is an Achilles heel of autocorrelation technique, was overcome when cross-correlation was applied that paved the way for measuring rotational structures such as tip vortices trailing from rotor blades. These advances provided sufficient measurement fidelity to study highly unsteady flows as well, such as the aperiodicity of wing tip vortices (Vogt et al. 1996). At the same time, the fully digital evaluation and increasing computational power facilitated new evaluation methods such as multi–pass or flow adaptive interrogation schemes. With the advent of full–frame interline transfer cameras, the first successful PIV measurements of the flow on fully articulated rotors were conducted in order to compare the evolving technique with LDV and measure rotor wakes in the context of blade-vortex interactions (Raffel et al. 1996a,b, Murashige et al. 1998).

Shortly after the first progressive scan camera PIV measurements, stereoscopic PIV was successfully employed to study the wake of hovering rotors (Heineck et al. 2000, Martin et al. 2000, McAlister et al. 2002) and articulated model rotors in wind tunnels (Raffel et al. 1998b).

4. DYNAMIC STALL AND REVERSE FLOW

The phenomenon of dynamic stall (DS) was first studied experimentally by Kramer (1932), who detected an increase in maximum lift at rapidly varying incidence angles due to gusts. This remarkable early experimental investigation has been made by means of hotwire anemometry and a piezoelectric balance. In the decades that followed, the model investigations of DS became more complex and mostly involved oscillating airfoils in wind tunnels (McCroskey and Fischer 1972). The majority of measurements involved interpreting results obtained by dynamic pressure sensors, hotwires, and dye flow visualization techniques, often independently. The augmented lift after the static stall angle (estimated from pressure measurements) was often assumed to have its source in the leading edge vortex. A verification of this assumption required accurate determination of the strength of the leading edge vortex and its size relative to chord and was sought to enable the optimization of CFD simulations. The ability of PIV to measure the velocity field over a plane at a given instant enabled such measurements.

Dynamic stall has been studied using PIV on 2D airfoils, 3D finite wings, and in model-scale helicopter rotors. The first application of PIV to an oscillating 2D airfoil was published by Shih et al. (1992), based on a water tunnel experiment at a low Reynolds number of 5000. In 1993, Raffel and Kompenhans applied PIV to study the unsteady flow field around a statically stalled airfoil at a Reynolds number of 3.4×10⁶. The first PIV investigation of DS at a Reynolds number of 4×10⁶ was carried out by Geissler and Raffel (1993) on a pitching 2D airfoil in a wind tunnel, see Fig. 5. Geissler and Raffel (1993) combined numerical and PIV data to study the effect of airfoil deformation on DS. Further PIV investigations of DS at similar Reynolds numbers (4×10⁶ to 6.2×10⁶) were carried out by Crisler et al. (1994), Geissler et al. (1994), and Raffel et al. (1995a,b) on pitching 2D airfoils in wind tunnels. The corresponding articles focused on the technical aspects of PIV and demonstrated the potential of instantaneous whole-field velocity measurements for the investigation of DS. In the following years, Wernert et al. (1994–1997) expanded the application of PIV to the complete phase-resolved DS cycle. The results of these studies on oscillating 2D wings comprise velocity, streamline, and vorticity information above deeply stalled airfoils. Wernert et al. (1994–1997) further described the formation of the DS vortex from multiple small-scale vortex structures and a strong aperiodicity of the flow field during stall, leading to the conclusion that point-based LDV measurements were unsuitable for measurements within the separated flow region of a dynamically stalled airfoil. A later study of Raffel et al. (2006) also investigated the onset of DS and the formation of a separation bubble by means of PIV and numerical simulations. The effort to characterize the many different aspect of DS is ongoing (e.g. Zanotti and Gibertini 2014, Davidson et al. 2015) and includes more and more complex flow scenarios like pitching airfoil aerodynamics during blade vortex interaction (Gibertini et al. 2014) and free stream Mach oscillations (Hird et al. 2015).

Fig. 5:

Instantaneous velocity field and vorticity contour above a dynamically stalled airfoil, after Geissler and Raffel (1993)

Multiple wind tunnel tests were conducted by Geissler et al. (2005) and Mai et al. (2006, 2008) to study the influence of leading edge vortex generators (LEVoGs) on the DS vortex. The application of PIV in these studies gave valuable insights into the flow field over the stalled airfoil and enabled the investigation of the mode of operation of the LEVoGs. This knowledge was expanded by Heine (2012) and Heine et al. (2009, 2011, 2013), who found that the disturbance generators increase the convection of vorticity and reduce the coherence of the DS vortex. Heine et al. (2011) also applied tomographic PIV to unravel the flow field around isolated LEVoGs under static flow conditions, shown in Fig. 6. The 3D measurement domain had a volume of 67 mm × 91 mm × 8 mm and resulted in 3.26 million vectors recorded simultaneously. Based on the volumetric velocity field information, the size and rotational direction of the pair of counter-rotating wake vortices was determined and their mutual interaction was analyzed (Heine et al. 2012). Joubert et al. (2012, 2013) studied a similar, but actively deployable, leading edge actuator on a pitching 2D airfoil by PIV measurements and URANS simulations. The effectiveness of the actuators was demonstrated by a highly reduced stall region above the airfoil. In another control study, PIV measurements were conducted by Gerontakos and Lee (2007) to investigate the effects of a trailing edge strip on the flow field of a dynamically pitching airfoil. In this study, Gerontakos and Lee (2007) found that a strip attached to the pressure side trailing edge (Gurney flap) increased the lift, drag, and pitching moment, whereas a strip on the suction side reduced all three coefficients. In a second investigation, Gerontakos and Lee (2008) applied PIV to study the effect of a trailing edge flap on the DS vortex, revealing a moderate influence on the leading edge vortex strength and no influence on its growth or detachment. Another attempt at DS suppression was made by Matalanis et al. (2016), who employed PIV to study the effect of combustion-powered blowing on DS.

Fig. 6:

Time-averaged wake behind a leading edge vortex generator (LEVoG). Iso-surfaces of helicity indicate the position of the main vortex and the transverse velocity component is color coded, from Heine (2011)

The studies of Mulleners et al. (2009) and Mulleners and Raffel (2011, 2013) focused on a detailed analysis of the DS vortex by applying PIV measurements with both a high spatial and temporal resolution to a pitching 2D airfoil. A uniqueness of this experiment is the simultaneous measurement of velocity field and dynamic surface pressure, which allowed relating the pressure signals to the flow features on the airfoil surface. These PIV investigations enabled a close examination of the formation and composition of the DS vortex. Mulleners and Raffel (2011, 2013) identified various coherent structures within the DS vortex and in the mixing layer between the separated flow and the surrounding flow field, as depicted in Fig. 7. They also determined the trajectories of these structures over the airfoil and unraveled the corresponding flow field by a proper orthogonal decomposition (POD) analysis.

Fig. 7:

Pitching 2D airfoil during upstroke with instantaneous flow field and detected vortex cores: (orange circle) clockwise and (blue circle) anticlockwise rotation; and color coded dimensionless out-of-plane vorticity component, after Mulleners and Raffel (2011)

Recently, Ramasamy et al. (2016) applied a POD analysis as an alternative solution to phase averaging. The phase average of aperiodic flow fields is not representative for the instantaneous velocity distribution and thus cannot be used for the validation of CFD predictions. To find an alternative solution, Ramasamy et al. (2016) conducted PIV and dynamic pressure measurements for a range of test cases on a 2D airfoil. Application of POD on PIV vector fields (with modes sorted based on turbulent kinetic energy) allowed identifying coherent structures that dominate the apparently random variation from cycle-to-cycle. The resulting mode shapes were potential alternatives to phase-average data for the validation of CFD simulations.

Apart from the majority of 2D DS investigations, there are a number of studies that strive to explore the 3D nature of DS. Yang et al. (2006) and DiOttavvio et al. (2008) for example induced DS on a model rotor by obstructing part of the rotor disk with a plate, therefore causing a rapid and dynamic change of the effective angle of attack on the blades. This type of setup allowed for the study of the radial flow velocity by PIV on the blades. A first investigation of DS using PIV on a model rotor under forward flight conditions was described by Mulleners et al. (2012). During the large-scale wind tunnel test, 2D velocity fields were measured at three radial blade positions. This data set allowed for the study of the 3D shape of the DS vortex, which was found to be aligned with the leading edge inboard but becomes more three-dimensional outboard where it interacts with the tip vortex . Recent studies by Raghav et al. (2013, 2014) investigated DS with a similar setup of a model rotor in a wind tunnel, including PIV measurements at various radial positions. Based on the acquired data, Raghav et al. (2013, 2014) determined the radial course of the separation line and found that the DS vortex had a closer proximity to the blade inboard, and a more coherent structure outboard. Cycle-to-cycle variations of the DS vortex were found to be relatively small, but the radial location varied, thus causing cycle-to-cycle variations of the pitching moment.

The 3D nature of the DS vortex also inspired wind tunnel investigations of pitching half-wing models with a free blade tip. The cut-off blades enable the study of interactions between the DS and tip vortices in a non-rotating frame of reference. Le Pape et al. (2007b) conducted a series of comprehensive wind tunnel tests on different airfoils including such a half-wing model. During these tests, Le Pape et al. (2007b) acquired hot film and surface pressure data and measured the velocity fields in the chord-wise and span-wise direction with LDV and PIV to investigate the onset of DS, influence of Reynolds number effects on the loads, and interaction between the DS and tip vortex. A similar experiment of Merz et al. (2014, 2015) combined pressure, blade deformation, and stereoscopic PIV-based velocity measurements at three span-wise positions to study the flow over a pitching half-wing model with a parabolic tip. The stereoscopic PIV data enabled the analysis of the span-wise onset of DS and the dynamically changing, strong radial flow components. The influence of DS on the wing tip vortex of this model was further studied by Wolf et al. (2015) through PIV measurements in the wake of the pitching airfoil. In this study, Wolf et al. (2015) analyzed the variation of the vortex trajectories, tangential velocity distribution, and vortex circulation over the pitching cycle and found that the tip vortex became less coherent during deep stall.

Finally, Nati et al. (2015) demonstrated the application of tomographic PIV for the study of vortical structures on a pitching 2D airfoil with a Reynolds number of 30,000. The PIV measurements allowed for the detailed investigation of a laminar separation bubble on the airfoil and the temporal development of 3D vortical flow structures above the airfoil.

Recent tests also focused on the reverse flow occurring on the inner parts of the helicopter rotor, where the forward flight speed of a helicopter exceeds the angular velocity of the retreating blade. Various airfoils in pitching motion have been investigated during forward and reverse flow with two-component high-speed PIV (Lind et al. 2014, Smith et al. 2016).

The same phenomenon has also been investigated by stereoscopic PIV measurements on the rotating rotor blade for different advance ratios, radial, and azimuthal positions (Hiremath et al. 2015 and 2016).

5. ROTOR WAKE INVESTIGATIONS

A large number of experiments have been conducted over the past decades in order to advance the understanding of the complex flow field around (model) helicopter rotors. Tip vortices are the strongest flow features in the rotor wake and are of great interest because of their influence on the induced flow field that directly affects the induced efficiency of the rotor. The characterization of tip vortices, i.e., their spatial location relative to the blade, core size, peak swirl velocity, vorticity, and circulation, is essential for developing strategies to alleviate their adverse effects and for generating mathematical models to predict rotor air loads. However, accurate estimation of tip vortices is not a simple task because of their inherent aperiodicity. PIV played a dominant role in the accurate measurement of tip vortex characteristics. A brief overview of other measurement techniques used historically in the rotor flow field is given before outlining the detailed contributions of PIV.

The first studies on helicopter blade tip vortices were conducted using whole-field smoke flow visualization (e.g. Gray 1957, Landgrebe 1971) and by seeding the tip vortex with smoke (Spencer 1970). Quantitative flow investigations on rotors using intrusive measurement techniques started in the 1960s. In the early 1970s, non-invasive, optical measurements (Laser Doppler Velocimetry - LDV) played a dominant role. Scully & Sullivan (1972) were among the first to apply LDV on hovering model rotors. In 1988, Thompson et al. made highly resolved measurements of the swirl velocity profile and core radius of a vortex shed from a single-bladed hovering rotor. With the development of lasers, flow visualization via laser light sheet illumination arose. This technique was applied e.g. by Leighty et al. (1991) and Ghee & Elliott (1992, 1995) for the detection of vortex trajectories in hover and forward flight conditions. Further improvements of the LDV technique allowed for finer resolution and the determination of a three component (3C) velocity field around the blade tip vortex (McAlister et al. 1995, Boutier et al. 1996). Studies like the one published by Devenport et al. (1996) found that the accuracy of LDV was limited by a random variation of the tip vortex positions, also referred to as vortex meandering or aperiodicity. To circumvent this limitation, research groups led by Leishman and others tried to minimize the meandering of hovering rotors by reducing the number of rotor blades to one, by preventing recirculation and wall effects, and by ensuring low turbulence intensity in the rotor inflow (e.g. Leishman et al. 1996, Han et al. 1997). These and other LDV studies focused on the detailed measurement of the peak values and temporal developments of the axial and swirl velocity components, as well as the core radius and circulation (Coyne et al. 1997, Bhagwat & Leishman 1998, Martin et al. 2000). This experimental data resulted in the formulation and improvement of various mathematical models for the velocity distribution around the tip vortices.

The development of Particle Image Velocimetry (PIV) enabled a second approach to dealing with the aperiodic vortex movement by instantaneously acquiring the entire flow field around the vortex. This allowed two kinds of averaging. The first approach is ensemble averaging, where the instantaneous vector fields are moved relative to each other so that the center of vortex coincides on all frames. The vector fields are then averaged to get an averaged flow field and mean vortex properties. The second approach, which is only possible using PIV, involves determining the vortex properties from instantaneous vector fields followed by estimating statistical quantities of the resulting data set. This not only provides mean vortex core properties but also a measure of the standard deviation that is essential to show magnitude of variation over time.

Saripalli (1995), Raffel et al. (1996b), and Murashige et al. (1997) were among the first to successfully apply PIV on helicopter rotors. Studies from Raffel et al. (1998a) and Ramasamy & Leishman (2006) compared results from simultaneous PIV and LDV measurements and promoted the validation of PIV as a reliable tool for blade tip vortex investigations. A number of publications about PIV on rotors focused on technical aspects and the challenges involved with measuring increasingly larger rotors during simulated forward flight (see e.g. Raffel et al. 1998b, Raffel et al. 2001, Richard & Raffel 2002, Raffel et al. 2004, Kindler et al. 2009, Norman et al. 2011).

The Higher harmonic pitch control Aeroacoustic Rotor Tests (HART I & II) represent two milestones of LDV and PIV application on rotorcraft. HART I included LDV measurements for the characterization of blade tip vortices (Splettstoesser et al. 1997). The HART II test comprised stereoscopic PIV measurements that resulted in vortex core radius and swirl distribution data in different parts of the rotor plane (Burley et al. 2002, Raffel et al. 2004, van der Wall & Richard 2006, Burley et al. 2006). A view of the HART II rotor with superimposed stereoscopic PIV images is given in Fig. 8. The rotor was operated in forward flight with a free stream velocity of 33 m/s. The accumulation of 2D measurement planes illustrates the vortex movement and decay with time as the vortices convect along the rotor plane. The HART projects were succeeded by the HOTIS (Hover Tip Vortex Structure) test, which aimed at complementing the HART II data set with comprehensive hover tip vortex data (Richard et al. 2006a,b, van der Wall & Richard 2008). High resolution 2C and 3C PIV measurements were conducted on the HART rotor in the rotor preparation hall at DLR Brunswick. The test included an assessment of the influences of interrogation window size and overlap on the vortex parameters. The results of the test further comprised blade tip deflections, vortex trajectories, an analysis of vortex meandering, and – for the first time – the combination of several velocity fields for the 3D reconstruction of a vortex segment. Other notable applications of PIV on rotors in wind tunnel tests include tip vortex characterizations by Yamauchi et al. (1999) and Kato et al. (2003).

Fig. 8:

HART II rotor in forward flight with multiple superimposed stereoscopic PIV images illustrating the vortex convection through the rotor plane

The first, full-scale wind tunnel tests on a helicopter rotor were conducted by Wadcock et al. (2011) and Yamauchi et al. (2012) using stereo PIV measurements for the localization and characterization of vortices on the advancing blade side. Flow field measurements covering 4.3- by 0.9-m were made at several thrust conditions. The processed data set was corrected for aperiodicity and the vortex core properties were estimated from instantaneous vector fields by simultaneously solving several vortices together (Ramasamy et al 2011).

In 2000, Heineck et al. tested a rotor in hover and applied PIV for the localization of vortices and characterized their wandering. McAlister et al. (2001a, 2003) applied stereoscopic PIV on similar rotors and investigated the temporal development of the core radius and peak swirl velocity. In these publications, McAlister et al. did not report clear trends for young vortices with a wake age of ψ_v < 30°. Ramasamy (2009) found similar inconsistent trends for vortex ages of ψ_v < 15°, indicating ongoing vortex roll-up. Ramasamy et al. conducted two PIV experiments to investigate blade tip vortices on a hovering model rotor in 2010 and 2011. The first experiment focused on the comparison of an untwisted and a highly twisted rotor and showed the differences in circulation distribution, vortex strength, and the flow field behind the entire blade. The second experiment was a high-resolution investigation of the vortices and focused on the swirl velocity profiles and the influence of the seeding void. Milluzzo & Leishman (2013) performed a hover test on a model rotor and focused on the tracking and characterization of vortices up to a vortex age of ψ_v = 420°. Similar techniques have been used by Kuerbitz and Milluzzo (2015) to study the effect of centrifugal pumping on rotor blade tip vortex diffusion. Other applications of PIV in rotor flow fields include wake measurements for deriving the circulation distribution and sectional drag along the rotor blade span. These measurements were integrated over the blade span, providing rotor thrust and power (or torque), respectively. The global force values were found to correlate well with thrust and torque sensor measurements (Ramasamy et al 2008), suggesting that PIV can be used to estimate loading distributions when the application of surface pressure sensors on rotor blades is not possible.

Recently, an investigation of active blade control by active twist actuation was conducted within the Smart Twisting Active Rotor (STAR) project. Bauknecht et al. (2015) and Bauknecht (2016) employed time-resolved stereoscopic PIV to track and characterize young vortices shed by a large-scale active rotor similar to the HART and HOTIS tests. The velocity data obtained was combined with reconstructed density information based on simultaneously performed time-resolved background-oriented schlieren (BOS) measurements, which enabled a thorough analysis of the vortex formation process. The high temporal and spatial resolution of the PIV data also enabled a 3D visualization of the tip vortex and vorticity sheet behind the rotor blade (Fig. 9) as a combination of time-resolved velocity and vorticity fields.

Fig. 9:

3D visualization of an instantaneous flow field behind the rotor blade. Isosurfaces mark the positions of the tip vortex and vorticity sheet behind the blade, from Bauknecht (2016)

Alongside measuring mean vortex properties, PIV experiments have been conducted to quantify turbulence inside the tip vortices trailing from a model-scale rotor (Ramasamy et al 2009). Shear stress and strain rate distribution were measured inside tip vortices using dual-plane PIV and were compared with fixed-wing measurements. The correlation was excellent suggesting PIV can be confidently used for turbulence measurements as well.

In addition to these model rotor tests, Raffel et al. (2001) and Kindler et al. (2009, 2011) performed PIV measurements on the tip vortices of a BO 105 helicopter under hover conditions. They investigated the swirl velocity profiles, vortex meandering, and vortex asymmetry and successfully demonstrated the first applications of PIV on a real helicopter.

The application of PIV to rotor flow continues to shed light on the complex aerodynamic mechanisms of the creation, convection, interaction, and break-down of blade tip vortices. The studies presented above also illustrate some of the challenging technical aspects of flow field measurements on model rotors. The trailing vortices of rotors are among the strongest that can be produced, and the associated high velocity gradients and seeding voids push modern PIV grid deformation methods to their limits. The large cycle-to-cycle variations of the blade and vortex motion necessitate elaborate aperiodicity corrections and precise azimuthal triggering of the measurements. Finally, the investigation of undisturbed vortices implies the elimination of wake recirculation and ground effects – a requirement that often conflicts with the available test facilities and safety considerations.

6. ROTOR IN GROUND EFFECT (IGE)

In the vicinity of the ground, trailing wakes from helicopter rotors interact with the ground during both hover and low speed forward flight. The complex interactions significantly affect several key aspects of helicopter operation. For example, the performance of the helicopter is affected during operation within about one rotor diameter from the ground, as the trailing wake-ground interaction changes the inflow velocities at the rotor disc. Consequently, this interaction results in improved performance in hover in IGE (Betz 1937, Knight & Hefner 1941) and reduced performance in low speed forward flight compared with out of ground effect (OGE) conditions (Sheridan & Weisner 1977).

The downwash from a hovering rotor IGE can also induce brownout conditions. Brownout causes erosion of engine blades and aerodynamic components such as wing surfaces, and increases the pilot work load because of the unsteady nature of the wake-ground interaction.

The flow field of a rotor IGE is a fundamental aspect of the helicopter operation. However, its prediction still proves to be challenging for both computational simulations and comprehensive codes. This is mainly due to the poor understanding of the interaction between the rotor wake and ground that can be directly attributed to the limited experiments available. As noted by Nathan and Green (2009), the dearth of high-fidelity measurements of the rotor flow field IGE comes from the difficulty to replicate the rotor IGE environment, especially for forward flight conditions.

The interaction of the rotor wake with the ground is an unsteady phenomenon that requires the simultaneous measurement of the entire flow field to characterize the flow features. Particle image velocimetry is an ideal technique for such studies as velocity information is acquired over the entire region of interest at any given instant. The successful application of PIV has enabled an improved understanding of helicopter rotor flow fields IGE both in hover and forward flight conditions. The following section provides examples for the application of PIV on rotor operating IGE that improved the understanding of the underlying flow features.

Tip vortices and, in general, the rotor wake are inherently aperiodic. As a result, the spatial locations of the tip vortices change slightly with respect to the rotor blade tip in successive revolutions (often within 1 diameter of the vortex core). The vortex characteristics acquired by point measurement techniques such as hot-wire anemometry or LDV thus exhibit aperiodicity-induced smoothing, which has to be corrected by a post processing procedure (Ramasamy & Leishman 2004). At older wake ages such as those found in IGE conditions, however, the flow exhibits a highly increased unsteadiness, making it uncorrectable. In addition to the inherent unsteadiness of the flow field, an unsteady pairing of tip vortices occurs (Lee et al. 2008). Understanding the process of vortex pairing and the resulting local velocity variations are only possible through the simultaneous acquisition of velocity information for an entire measurement plane.

The quantification of unsteady flow phenomena is particularly useful for the investigation of brownout as the vortex pairing process has been hypothesized to provide the necessary energy to uplift the sand particles from the ground (Johnson et al. 2009). Phase-locked low-speed and high-speed PIV measurements (Johnson et al. 2010a) of model-scale rotor flow fields showed vortex pairing and the resulting fluctuations in local velocities. Dual-phase PIV measurements for model scale rotors hovering IGE above a sand bed illustrated how the sand particles were uplifted during vortex pairing, especially in a region close to the tip of the rotor (r/D=1). Attempts to reduce the strength of outwash velocities of rotors IGE have been conducted using rotor blades with various tip shapes (Milluzzo et al. 2010). The interaction of tip vortices with the ground produces a counter rotating vortex pair that is believed to substantially contribute to the brownout phenomenon through the ejection of particles into the flow field (“wall jet ejection”). The identification of the unsteady counter-rotating vortex was only made possible when the entire flow field was captured instantaneously using PIV (Johnson et al. 2010a). Further investigations on the influence of turbulence on brownout were conducted using time resolved dual phase PIV as well (Rauleder & Leishman 2013).

Recently, flow field measurements with a multi-camera PIV system were conducted on a model rotor operating IGE. The images were stitched after the test, resulting in a flow field spanning approximately four rotor diameters from the rotor axis (Ramasamy et al. 2015). The resulting instantaneous flow fields (Fig. 10) revealed the presence of large-scale shear layer vortices of the size of the rotor height in the flow field. The experiment also compared the instantaneous flow field of an impinging jet with the rotor flow field using the same experimental set up. The comparison confirmed that wall jet characteristic variables are applicable to rotor outwash. Instantaneous flow fields also revealed similarities and differences between the flow fields of a tandem and single rotor configuration IGE (Ramasamy 2015). The interaction between the tandem front and aft rotor trailing wakes, followed by their subsequent interactions with the ground, produced wall jets with different evolutionary characteristics (in terms of wall jet width and peak outwash velocity) than the single rotor flow field.

Fig. 10:

Rotor in ground effect showing large shear layer vortices

For low speed forward flight IGE, helicopters exhibit increased power requirements compared to hover IGE, sometimes as high as forward flight OGE. The increased power requirement to produce the same thrust can be attributed to the wake recirculation and the formation of a ground vortex, which increases the inflow velocity at the rotor disc. The effect of the ground slowly reduces with increasing forward flight speed, and completely vanishes around µ=0.1. Early experiments studying low speed forward flight IGE used hotwires to measure the flow field. Forward flight was simulated in four different ways: (1) by full-scale flight tests (Kutz et al. 2013), (2) by towing model-scale rotor (Curtiss et al. 1984) (3) through wind tunnel tests with a fixed ground (Boer et al. 2001, Bourne et al. 2014), and (4) through wind tunnel tests with a moving ground (Nathan & Green 2009).

The experiments by Curtiss et al. (1984) on low speed forward flight classified IGE flow conditions into two states, as shown in Fig. 11: (1) a recirculation state and (2) a ground vortex state. A clear distinction between the two states was made through the evaluation of the rotor wake. The recirculation state is characterized by the downwash from the rotor, which features an initial contraction followed by an expansion as it approaches the ground and a subsequent recirculation into the rotor plane. The ground vortex state occurs at slightly higher forward velocities where a more concentrated vortex forms below the front edge of the rotor disk. No recirculation occurs in this situation as the outer part of the blade sees upwash while the inner part experiences downwash. The presence of a ground vortex introduces lateral cyclic disturbances that require compensatory control inputs and thus increase the pilot’s work load. The size and strength of the wake recirculation, ground vortex, and their evolution significantly affect the control/handling qualities of the helicopter as well as the brownout phenomenon.

Fig. 11:

Schematic of rotor wash in low speed forward flight showing recirculation and ground vortex zones, from Nathan and Green (2009)

Various studies reported contradictory findings for the characterization of forward flight IGE flow fields, e.g. for the identification of the threshold velocity that separates the recirculation and ground vortex states and the amount of observed wake unsteadiness for each state. Nathan & Green (2009) conducted a series of measurements using two model-scale rotors and a wind tunnel with a ground plane capable of moving at free stream velocity. The results of this investigation demonstrated the significance of a moving ground plane as opposed to a fixed ground for the state of the rotor wake. Representative PIV flow fields corresponding to two different forward-flight speeds are shown in Fig. 12 and exhibit a significant difference of the ground vortex despite the relatively small difference in advance ratio. Through comprehensive PIV measurements on the rotor flow field in forward flight IGE, Nathan & Green (2009) were able to develop criteria that allowed separating recirculation and ground vortex zones decisively. The relation between unsteadiness found in the wake (recirculation region), lateral control inputs, and vorticity present in the wake was thus shown by means of stereoscopic PIV measurements.

Fig. 12:

Effect of forward flight speed on the ground vortex (a) μ*=0.67 (b) μ*=0.7, from Nathan and Green (2009)

7. ROTOR WAKE INTERACTIONS WITH OBSTACLES

In addition to using PIV to quantify the detailed structure and behavior of rotor flow field constituents such as blade tip vortices, shear layers, and leading/trailing edge vortices during stall, PIV has been used to reveal the impact of the entire rotor wake interacting with the fuselage, other aircraft, and building structures. Understanding and quantifying these global wake interactions is important when assessing the operational impact of rotorcraft on the surroundings. Also, the location and strength of the main rotor wake interaction with fuselage components such as the tail boom and tail rotor can drive the vehicle design process. The following section provides examples of how PIV was used to quantify the global behavior of rotor wakes.

Barbagallo et al. (2000) demonstrated the applicability of the PIV technique to study rotor wake interactions on a representative model-scale helicopter under the HELIFLOW project. The experiment, conducted in the 24-ft Farnborough wind tunnel, used hotwires and PIV to study main rotor wake interactions with the empennage and horizontal tail planes. The PIV (vertical plane, 0.15-m high by 0.18-m wide) and hotwire measurements were used to characterize the pitch-up behavior in low-speed flight experienced by helicopters with tail planes. The work under the HELIFLOW project, which also included the investigations by De Gregorio (1999) and De Gregorio and Tino (2000), demonstrated the effectiveness of PIV for identifying and quantifying wake interaction phenomena, especially compared to point-measurement techniques.

Wadcock et al. (2004) and Silva et al. (2004, 2005) used 1/48th-scale models of three different rotorcraft and an amphibious assault ship to investigate the aerodynamic interaction between the ship, a V −22 tiltrotor, a tandem rotor helicopter, and a single main rotor helicopter, see Fig. 13. The test was conducted in the Army 7-ft by 10-ft Wind Tunnel at NASA Ames Research Center with a setup consisting of a model scale ship, an upwind tandem rotor helicopter on a sting mount, and an on-deck V-22. A primary test objective was the quantification of the flow field disturbance generated by the combined wake of the ship and a rotorcraft operating upwind of a V-22 on-deck. PIV images were acquired in a vertical cross-flow plane approximately 1.83-m wide and 1-m high. Figure 13 (from Wadcock et al. 2004), shows the velocity field at the location of the on-deck V-22 for the specified upwind aircraft heights above the deck (WHOD) with the ship yawed 15° to starboard. The velocity fields clearly show regions of asymmetric upwash and downwash in the vicinity of the on-deck V-22 rotors, thus inducing a roll response by the V-22. The velocity fields together with aircraft force and moment data provided improved understanding of the aerodynamics driving the ROD phenomenon. Nacakli et al. (2012) performed a more fundamental study of rotor wake/ship interaction. In this small-scale study, the position of an isolated rotor was varied downstream of a backward-facing step (representing a simple ship model). Velocity fields, measured downstream of the step using PIV, characterized the changes in flow behavior and rotor thrust as the rotor position was varied with respect to the step.

Fig. 13:

Model scale V-22/Ship/Helicopter air wake Interaction test (from Wadcock et al. 2004). Velocity field at landing Spot 7 with CH-46 at Spot 6 for wind-over-deck = 345° at 35 knots full-scale. WHOD = full-scale wheel height above deck. View is looking upstream, every other vector shown

The results from the small-scale V-22/ship testing prompted additional investigations of tiltrotor wake interactions, specifically, the interaction of two tiltrotors in formation flight. Romander et al. (2006) used aircraft force and moment data augmented by PIV measurements of the velocity field to study the impact of the wake from the lead aircraft on the trailing aircraft. The PIV measurements showed that the lead aircraft wake persists far downstream. The vertical position of the wake is dependent on descent angle and flight speed, so lateral separation was determined to be the best means of avoiding adverse interactions between aircraft.

Using a 1/7th-scale model of a Dauphin 365N helicopter (fuselage and rotor), Le Pape et al. (2007a) surveyed the flow field around the helicopter at ten different planes using PIV in the ONERA F1 wind tunnel. The planes were oriented normal and parallel to the freestream velocity, resulting in the capture of the super-vortices from the edges of the rotor disk and determination of the wake skew angle – important features for characterizing the interaction between the wake and fuselage.

Jenkins et al. (2009) quantified the aerodynamic interaction between the wake of a main rotor (4 blades, 5.89-ft diameter) and other parts of a model-scale compound helicopter using two stereo-PIV systems. The model was installed in the NASA Langley 14-ft by 22-ft Wind Tunnel (see Fig. 14). Figure 14 shows an example of the feature-rich measured flow field.

Fig. 14:

Aerodynamic interaction tests of a model compound helicopter (from Jenkins et al. 2009). a) Installation in the NASA Langley 14- by 22 Ft Wind Tunnel, b) streamwise vorticity (view looking upstream)

The European Union GOAHEAD project generated perhaps the most comprehensive aerodynamic database to date for a complete helicopter configuration (see Raffel et al. 2010). A model NH90 fuselage with an ONERA 7AD 4-bladed rotor was tested in the DNW-LLF facility (model and rotor were 1:3.88 scale). Extensive PIV measurements, described in De Gregorio et al. (2012), were carried out using five lasers and four cameras, see Fig. 15. Velocity fields were acquired at 37 measurement locations, including above and below the empennage, above the horizontal tail, and at various radial stations on the advancing and retreating rotor blades. Comparisons of the flow field downstream of the rotor hub, with and without the rotor, quantified the influence of the rotor wake on the tail surfaces and Empennage (see Fig. 15).

Fig. 15:

PIV measurements for the GOHEAD test in the DNW-LLF depicting the flow field over the tail boom (from DeGregorio et al. 2012)

Droandi et al. (2013, 2016) used PIV to study the rotor-wing interaction for different wing positions of a half-span tilt-wing configuration in hover. Flow field planes along the wingspan were measured using PIV; the rotor wake boundary was also determined. These types of flow field information must be considered when sizing the wing of a tilt-wing or tiltrotor aircraft.

The main rotor wake can also influence the effectiveness of devices installed on the fuselage and empennage to improve aerodynamic efficiency. For example, using PIV to survey the flow field behind a fuselage ramp, Schaeffler et al. (2016) determined that the actively-controlled synthetic jets embedded in the ramp altered the rotor wake shape and trajectory. The altered trajectory of the wake changed the surface pressure distribution on the ramp, which resulted in reduced fuselage drag.

The rotorcraft community has effectively used the PIV technique to characterize the global rotor flow field. Using measured flow fields for a range of test conditions and separation distances between the rotor(s) and fuselage/empennage/wing components, informed design choices for locating horizontal tails, vertical tails, wings, and flow effectors can be made to avoid adverse interaction with the main rotor wake.

7. CONCLUSIONS

The state of the art of PIV applications for helicopter rotor aerodynamics has been summarized in this review paper. A comprehensive overview was given on PIV experiments that have been conducted between 1992 and today. These experiments focused on different domains of the flow field around the helicopter, such as dynamic stall, the rotor wake, rotor flow in ground effect, and rotor wake-obstacle interactions.

A brief introduction on the technical principle of PIV was followed by a summary of early helicopter flow field investigations with PIV. An overview of pre-PIV dynamic stall research was given next, followed by a description of early PIV investigations on dynamically stalling 2D airfoils. Over the past two decades, dynamic stall research featured a transition from 2D oscillating airfoil studies to investigations of 3D airfoils and dynamically stalling rotating wings. Concurrently, the PIV measurement technique became increasingly sophisticated and thus enabled highly detailed flow-field investigations and a massive gain of knowledge on the dynamic stall process.

A summary of rotor wake studies was given, with a focus on blade tip vortex investigations. Before the rise of PIV, this subfield of helicopter aerodynamics was largely dominated by LDV measurements. With increasing measurement resolution and reliability, PIV became the dominant measurement technique for blade tip vortex investigations. The strength and position of vortices was determined for different rotor configurations in hover and forward flight and even for full-scale rotors. PIV thus enabled the detailed investigation of the creation, convection, interaction, and break-down of blade tip vortices.

Further areas of PIV application include the interaction of the rotor wake with the ground. The basic principle of this interaction was explained, followed by a description of key experiments that investigated the large vortical structures below rotors in ground effect by means of PIV. Other PIV-based experiments focused on the phenomenon of brown-out and the underlying highly unsteady process of particle uplift. The capability of PIV to measure large and unsteady flow fields enabled investigations of the rotor-induced flow field on hovering rotors and rotors in forward flight in ground effect.

PIV studies on the interaction of the rotor wake with obstacles and other aircraft were summarized. Typical applications of PIV were illustrated, such as the investigation of a rotor wake interaction with another aircraft on top of a ship. Both conventional and unconventional helicopter configurations with lifting surfaces were studied in detail by means of PIV.

The vast number of reported rotor studies demonstrates the significance of PIV for the study of rotor aerodynamics. The capability to measure large-scale, whole-field, and non-intrusive velocity fields makes the PIV technique an ideal choice for studying the unsteady and complex flow field of helicopter rotors. Future applications of tomographic and stereoscopic PIV systems with high spatial and temporal resolution will further our understanding of complex processes such as blade-vortex interactions, 3D dynamic stall, and wake-fuselage interactions and thus contribute to the development of the next generation of helicopters.

LIST OF SYMBOLS

a	speed of sound	[m/s]
A	amplitude of oscillation	[°]
∆Ac	size of particle void in PIV	[m]
C1,2	correlation plane
c	chord length	[m]
CP	pressure coefficient
I1, I2	interrogation windows
Lw	interrogation window size	[m]
Ma	Mach number
r	radial coordinate	[m]
rc	vortex core radius	[m]
R	blade radius	[m]
Rev	vortex Reynolds number
u,v,w	velocity components	[m/s]
UM	in-plane velocity	[m/s]
Uoo	free-stream velocity	[m/s]
Vr	radial velocity component	[m/s]
Vz	axial velocity component	[m/s]
Vθ	swirl velocity component	[m/s]
x, y, z	Cartesian coordinates	[m]
α	angle of attack	[°]
êz	unit vector in z direction
Ψ	vortex age	[°]
ν	kinematic viscosity	[m²/s]
ω	vorticity	[s−1]
Ω	rotor rotational frequency	[s−1]
Γ	circulation	[m²/s]
Γ1(P)	scalar field at given point P