Radial Flow Measurements Downstream of Forced Dynamic Separation on a Rotor Blade

Jaesuk Yang* Balakrishnan Ganesh*, Narayanan Komerath†

Georgia Institute of Technology, Atlanta, Georgia, 30332

Dynamic flow separation is forced on the suction side of a rotor blade in a hover facility, to study the development of radial flow in the separated zone. Inflow to the rotor disc is obstructed with a plate to induce separation over part of the rotor azimuth. Pulsed laser sheet imaging and particle image velocimetry (PIV) are used to verify the occurrence of separation, and to capture features of the separated flow. Results verify the occurrence of stall in the obstructed-inflow sector, as well as the presence of lift elsewhere in the rotor disc, thus validating the forcing technique. In the stalled region, radial inflow is stopped near the rotor blade, but strong outward flow is not seen to develop. This appears to be due to the presence of separation cells along the radius.

Nomenclature

CT = thrust coefficient ! = angular velocity of the rotor R = rotor radius r = radial coordinate UR = radial velocity UT = tangential velocity

1. Introduction

Dynamic stall occurs on rotor blades in edgewise flight at high advance ratio. Cyclic pitch variation is used to balance the moments on the advancing and retreating sides of the rotor disc. The pitch increase on the retreating blade, and the rapid azimuthal change in freestream and inflow velocity, cause the effective angle of attack to exceed the static stall limit rapidly and substantially. This triggers a sharp dynamic lift increase followed by sudden loss of lift, and the stall persists until the blade is well under the static stall limit again. Dynamic stall is a first-order limiter of rotorcraft forward speed and ride quality, and causes large blade loads. As such, the phenomenon has been studied extensively. Much is known1 about two-dimensional dynamic stall of airfoils, both computationally and experimentally. Concepts for stall alleviation2 and elimination3 have been developed. Since rotor blades have high aspect ratio, fixed-wing studies with suitable yaw and spanwise lift distributions have been used as analogues for 3-D dynamic stall. Much less is known about the additional complexities on rotor blades due to the presence of strong radial acceleration. This is the aspect studied here. Shortly after the onset of dynamic stall on a retreating helicopter blade, the stall is established inboard, while the tip is still unstalled. The precise mechanism by which the actual dynamic stall initiates, involves compressibility and supersonic flow, as shown by Carr et al 4 in two-dimensional oscillating wing experiments. However, the region downstream of the separation line is certainly at low Mach number. The nature of the separation line on a rotating blade must involve interaction between the stall vortex as defined from 2-dimensional concepts, and the details of the separated flowfield downstream. Thus, study of this region opens the way to understanding 3-D dynamic stall. The motivation for the present studies is two-fold. The first is to study the radial flow phenomena during dynamic stall, as indicated above. The second is to do this in experiments that allow the development and exercise of hub-mounted instrumentation. Such instrumentation is needed to enable studies of flow details close to the blade surface in a continuous temporal sequence, as opposed to instrumentation mounted in the laboratory frame. The latter aspect drove the design of the experiment reported here.

* Graduate Research Assistant, School of Aerospace Engineering, MS0150 AIAA Student Member † Professor, School of Aerospace Engineering, AIAA Associate Fellow

36th AIAA Fluid Dynamics Conference and Exhibit5 - 8 June 2006, San Francisco, California

AIAA 2006-3377

Copyright © 2006 by J.Yang, B.Ganesh and N.Komerath. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.

2. Previous Work It is known that flow separation occurs on the retreating rotor blade at the edges of the performance envelope.

During maneuvers and operations at high altitudes, the rate of increase of angle of attack on the retreating blade side (RBS) is large enough to cause substantial lift increase past the static stall angle of attack, followed by sharply negative pitching moment and vibratory loads due to stall. Reattachment is delayed in a hysteresis loop which persists well into the next quadrant of the rotor disc. Efforts to capture this phenomenon have proceeded through 2-dimensional experiments on pitching airfoils in wind and water tunnels5,6 and numerical predictions proceeding from inviscid formulations to full Navier-Stokes simulations. Detailed diagnostic experiments on dynamic stall by Carr et al defined the role of compressibility in dynamic stall – and more recently developed techniques for delaying dynamic stall using surface modifications. In the early 1990s it was concluded that 2-D dynamic stall models were overpredicting the severity of flight results on helicopter blades. Research on 3-D dynamic stall showed that the stall initiation was a localized event, thus making the timing of the stall difficult to predict and control. For instance, Van Dommelen and Shen7 showed the development of a singularity at the onset of turbulent boundary layer separation, followed by “eruptions” of boundary layer fluid into the outer stream.

Lorber et al8 argue that controlling retreating-blade stall could enable a 23% increase in payload capability for utility helicopter missions, and a 35% increase in range and endurance with a 33% addition in fuel. Stall control to 85% of the blade radius can delay stall by 5 degrees, enabling a 10% thrust increase throughout the speed range from 40 knots. Control to 95% radius would enable a 10% increase in CLmax in addition to the 5-degree delay in stall angle, giving roughly an 18% thrust increase. A speed increase of roughly 20 knots was also predicted. Bousman9 points out that well-designed rotors avoid significant stall-related loads over most of the flight envelope. However, the limit on substantial portions of the steady and maneuver flight envelope is posed by blade stall9,10. The section Reynolds number varies widely over this range, and the blade angle-of- attack goes well past the static stall angle. Table 1 below gives typical operating parameters of a full-scale main rotor, under conditions where dynamic stall problems are typically cited.

Table 1: Summary of typical rotor blade parameters related to retreating blade stall9,10 Full scale radius 40 ft (12.192m) RPM 300 (31.416 rad/s) Blade chord 2 ft (0.6096m) radial r/R 0.2 Centrifugal acceleration, Gs 245.51 Section chord Reynolds number 4 million Relative external Mach number 0.3 - 0.5 Peak local Mach number 1.1 – 1.3 Typical chordwise location of separation, x/c 0.05 to 0.15 Typical cyclic pitch amplitude 10 - 15 deg. Non-dimensional maximum instantaneous pitch rate 0.002 to 0.02 Reduced frequency based on blade half-chord 0.05 to 0.15 The high relative Mach numbers encountered on the advancing blade side (ABS) dictate the use of thin airfoils with small nose radius and a relatively flat upper surface to reduce shock drag and impulsive noise. As these blades pitch up on the RBS, the flow over the nosetip encounters large negative pressure coefficients (Cp of –5.4 shown at 0.5% chord9, exacerbated by unsteady effects. Thus local Mach number reaches9 as high as 1.3. Carr and colleagues4 have shown the effects of shocks in the development of dynamic stall.

Influence of spanwise flow on rotating blades Bousman’s data10 show that moment stall begins at an early azimuth where the freestream has a substantial

spanwise inboard-directed velocity component, whereas close to the surface, radial acceleration drives the boundary layer fluid outward along the span. This must generate a strong velocity gradient which changes rapidly as the spanwise component comes to zero at 270 degrees azimuth. Thus, the correct fluid dynamics problem to simulate in experiments to capture stall initiation/suppression appears to be one of three-dimensional separation over a yawed wing with high radial acceleration at the surface. High aspect ratio is a secondary consideration here.

Influence of Radial Acceleration This has been debated in the literature some years back, but has generally not been given much attention due to the difficulty of conducting boundary layer measurements on rotating blades. Coton11 cites the substantial differences in the inboard pressure distribution on rotating wind turbine blades compared to 2-D models, preceding dynamic stall. He also cites the delay in forward movement of the separation region (prior to lift-off of the dynamic lift vortex) due to spanwise flow. Corten12 has used a “stall flag” method to capture the occurrence of stall on full-scale wind turbine blades. His analysis shows that in the dynamic stall region, there is a substantial radial pressure gradient and accompanying radial flow. Corten also cites laser velocimeter results13 showing the formation of vorticity directed parallel to the rotor axis along the aft portions of the blade. Radial acceleration must have a strong influence on the lift and pitching moment evolution after stall occurs. Xu14 shows that the influence of rotation (centrifugal effects in the boundary layer) on a wind turbine blade, when modeled using a full Navier-Stokes formulation, shows a substantial stall delay in the 3-D case compared to 2-D cases where there is no rotation effect.

Liou15 compared laser velocimeter results for the same blade operated as a rotor in the 9’ facility used in the present work, and as a fixed-wing with the appropriate inflow correction in a 7’ x 9’ low speed wind tunnel. We found, in cross-flow measurements made at the blade trailing edge, a substantial radial flow induced in the separated zone, compared to the fixed-wing measurements. The flow over the same blade was measured, in a 9’ rotor facility in “hover” at pitch angles of 15 deg. and 30 deg. and then as a fixed-wing in the John Harper Wind tunnel. The angle of attack in the wind tunnel was set to correspond to the pitch angle, corrected for the averaged inflow from the rotor experiment. Thus the angles of attack in the wind tunnel were substantially lower. These studies illustrated some interesting features. At the trailing-edge cross-flow plane, the radial velocity was drastically different between the rotating and fixed tests. The effect of the radial flow in the boundary layer was clearly visible. Over the blade tip, the tip vortex was seen to form and burst, much like on a delta wing .The region of separated flow included a large “dead-air zone” where the flow moved essentially at the same mean velocity as the blade.

Influence of Pitch Rate The influence of pitching can be broken up into that due to quasi-steady change in angle of attack, a vorticity convection effect, and the unsteady acceleration effect of the wall on the flow. The last of these is negligible at the low reduced-frequencies involved in the rotor cyclic pitch case. The “moving wall effect” is small in the rotorcraft case. In other words, an actual blade pitching rate is not essential to simulate the essential features of dynamic stall, if the effective angle of attack can be changed rapidly by other means.

3. Measurement Approach As discussed above, the objective is to study the flowfield behind the separation line on the inboard region of a rotating blade, and to do so in a setup where hub-mounted instrumentation can be integrated into the system gradually. The approach is to use a single-bladed rotor in a hover facility, and alter the effective angle of attack suddenly over part of the rotor disk. This is done by placing a plate in the inflow region, obstructing inflow over a sector of the inboard region of the rotor disk. The first issue is to check whether this approach actually causes stall in the obstructed region, and the next is to see what happens to the radial flow downstream of the separation line. The above concept for inducing stall has important advantages over going directly to the wind tunnel for forward-flight experiments. Firstly, conditions of zero spanwise freestream flow can be studied, while maintaining the radial acceleration and high pitch angle features of retreating-blade stall. Secondly, the tangential and radial extents of the region of stall can be varied, without changing the rpm.

Facility Description The experiment is conducted using a single-bladed rotor (Table 2) with a flap degree of freedom, in a hover facility.

The test cell is 2.74x2.74x6.4 m. The rotor is mounted on a horizontal shaft and is driven by an 11.2kW variable speed direct current motor. The speed can be varied continuously from 0 to 2000 RPM and is measured by electronically counting the pulsed output of a magnetic pickup which senses the passage of the teeth of a gear mounted on the drive shaft. The constant chord, untwisted blade has a square, flat tip. Its leading edge and spar are made of steel and the trailing edge is made of aluminum. The blade is statically balanced by a counterweight. The collective pitch can be manually adjusted, and the blade is free to flap within a small range of

Table 2: Rotor Blade Geometry Blade radius = 0.61m Blade chord = 0.13m Aspect ratio = 4.8 Airfoil section: NACA 0012 Pitch link range: 0 to 18 deg.

angles. A Hall-Effect sensor sends a pulse to an external signal generator every time a tab on the rotor shaft passes, providing azimuth reference. A plate in the shape of a 60-degree sector is mounted upstream of the rotor disc as shown in Figure 1, in order to restrict the inflow velocity over that part of the disc. The flow is seeded with atomized food-grade mineral oil used for laser velocimetry. The illumination source is a pair of pulsed Nd-YAG lasers, with coincident output beams, expanded into a laser sheet in the region of interest. A 10-bit Charged Couple Device (CCD) camera of 1360x1024 pixel resolution is used to obtain double exposed flow images for Particle Image Velocimetry (PIV). Signal-to-noise ratio begins to degrade for particle displacements exceeding 1/4th of the interrogation spot size. Measurements at 500 RPM were conducted with 50 microseconds pulse separation. Initial measurements were made in cross-sections at selected radial locations to look at the flow development over the airfoil section. To see the radial velocity, the laser sheet was aligned with the radius, perpendicular to the rotor disc. The viewing region was a 13.76 cm (radial coordinate) x 11.23 cm (axial coordinate) rectangle. The CCD camera was located 38 cm from the axis of rotation, as shown in Figure 2. To capture the axial flow, the camera was positioned on the ceiling of the facility. For this experiment, 90 degrees azimuth corresponds to 1.9 cm from the leading edge of the blade. At 500 RPM, an azimuth increment of one degree would take 0.000333 seconds at the blade tip. Then 0.000333 second was added for each increment in azimuth until 106 degrees azimuth, which is 2.54 cm from trailing edge. Radial velocity is captured at various locations along the spanwise viewing plane covering 13.7 cm of the blade about the 38 cm span as the center.

Figure 1. Inflow obstruction picture and sketch of the inflow region.

Figure 3. Coordinate system and chordwise and spanwise

measuring planes.

Spanwise section

Chordwise cross section

Axial flow, V

Tangential flow, UT

Radial flow, UR

Figure 2. Camera orientation viewed from the suction side (up stream)

4. Flow Measurements From previous experiments in this facility15, it is known that the blade used here operates with essentially attached flow at 15 degrees pitch, at 500 rpm (30 m/s tip speed). Initial experiments PIV measurements without the inflow obstructor showed that at a pitch setting of 15 degrees the rotor was producing thrust, and the flow over the suction side of the blade did not show stall. This was close enough to the limit of unstalled operation and beyond the blade static stall limit if the inflow were to be obstructed. Following this, the inflow obstruction was set up. Getting seeded flow to reach the desired measuring planes with the obstruction in place, proved to be quite difficult, and measurement in the chordwise/axial planes was deferred. Choice of a radial-axial measuring plane overcame the uncertainty in the radial station at which the seeded flow would pass. This measuring plane also provided access to the radial flow induced by the rotor blade and its wake. Most of the data presented here are from this plane. Qualitative Visualization With the obstruction installed, Pulsed Laser Sheet Imaging (PLSI) studies were conducted at 70 RPM, which is slow enough to capture successive video images, showing the same seeding during a single-event approach and passage of the blade through the seeding cloud, albeit at low Reynolds number. The PLSI showed that the flow across the disc (axial velocity component), which is usually sharply accelerated as a lifting blade passes, was being stopped suddenly, indicating blade stall. In Fig. 4, as the blade moves through the seeding cloud, very little axial flow movement can be seen (#1 and #2), showing the absence of strong suction. Soon after the blade passes through, the seeding resumes its downward movement. (#3). Viewed in detailed image sequences, this visualization made clear that the blade was essentially just a blunt body moving through the flowfield, and was not generating any significant lift. The inflow during the rest of the rotor cycle was due to the lift generated when the blade was beyond the inflow-obstruction region. This finding is quantified by PIV. Quantitative Visualization For PIV, four image sets of fifty image pairs were taken for each blade azimuth of the measuring plane. The spanwise–axial section was selected to capture the radial flow velocity (Figure 5). The measuring plane was illuminated along the span with a rectangular viewing region of 12.8 cm (spanwise) x 9.6 cm (axial). This was done in six two-degree increments of rotor azimuth spanning blade passage. To acquire the vector data a fine interrogation mesh was applied to the image sets. It took the computer around three hours to run a batch of four image sets. From this fine mesh a coarse 20 by 20 grid was made using TECPLOT plotting software in order to see the vectors without overlap. In TECPLOT, a multicolor vector coloring was used to show the magnitude of the inflow velocities.

5. Evidence of Blade Stall The experiment here is conducted under conditions where the aerodynamic load is small compared to the inertial loads. Also, no direct thrust measurement is made, and no pressure sensors are used. Thus, the occurrence of stall was verified by comparing the axial velocity induced by blade passage with the “background” level attributable to the tip vortex system and lift generation through the rest of the blade cycle. Previous measurements with a lift-generating rotor have shown that a very large inflow velocity spike occurs during blade passage, followed by settling to the “background” level very quickly for the rest of the cycle. However, if the blade is stalled as it passes the measuring location, and generating lift through much of the rest of the cycle, we should expect to see only the smaller perturbation in the inflow velocity during blade passage, associated with passage of a blunt body. It is expected that the flow will be pushed away by thickness of the blade, followed by a briefly increased inflow. Velocity results from the 70 rpm tests, followed by cleaner results at 500 rpm, verified that this was occurring. Figure 5 shows that at midspan, at 270 degrees azimuth (blade is at the opposite side of the rotor disc) the measured inflow velocity was about 2.5 m/s for the obstructed case and 3 m/s for the control case, as shown in Fig. 5. This is in line with rough expectations from Momentum Theory.

6. Radial Flow Figures 6 through 9 show contours of the radial velocity component, measured at various locations along the spanwise viewing plane covering 12.8 cm of the blade centered on the 38 cm radial location (midspan). At the leading edge plane, (Figure 6), without the obstruction, at 90 degrees azimuth, when the blade is assumed to be lifting and at the measurement region, the radial velocity component is in the range of 0.5 m/s (red region). However, when the blade is at 270 degrees (as far away as possible from the measuring region) the radial velocity is on the order of 1.4 m/s, as shown in Figure 7. This is the “background” level of radial velocity, attributed to the tip vortex.

Figure 4: Sequential Blade and Seeding Visualization and Schematic View

1

2

3

Seeding cloud

Laser sheet

0

0.5

1

1.5

2

2.5

3

3.5

4

0 0.2 0.4 0.6 0.8 1

Chord, x/c

Ax

ial

Vel

oci

ty,

m/s

Control

Obstructed

Figure 5. Axial Velocity Profile at Midspan

Comparing Figures 8 and 9 for the same cases as above, but with the obstruction present, we see that the radial velocity remains near the background level. Thus in the case with obstruction, the radial velocity above the blade leading edge region is not very different from the background.

This sets the context to view the change of radial velocity from leading edge to trailing edge over the blade, as it moves through the measurement plane, and compare the cases with and without obstruction. The measurements at mid-span are shown. In Figures 10 and 11, the blue lines show the deceleration from leading edge to quarter chord at 5 cm from the surface of blade, while the orange line is 10 cm from the surface. Figure 10 shows the radial velocity as a function of chord at mid-span without the obstruction. There is less discontinuity after the quarter chord than in the case with obstruction (Figure 11). The case with the obstruction, in Figure 11, shows a quick change in the magnitude of the radial velocity. The velocity fluctuation increases after the quarter chord. The reason why the

Figure 6. Radial velocity contour at leading edge, no obstruction

Figure 7. Radial velocity contour with blade at 270 degrees azimuth, no obstruction

Figure 8. Radial velocity contour at leading edge, obstructed case

Figure 9. Radial velocity contour with blade at 270 degrees azimuth, obstructed case

radial inflow velocity is initially higher in the obstructed case is that while the axial flow has been obstructed, the radial flow has not been obstructed. So when the pressure drop occurs near the upper surface, the radial inflow is accelerated as opposed to the controlled case where axial component of the flow would be sufficient in pressure recovery. The range of the radial inflow velocity variation for both cases (controlled and obstructed) is the same up to the quarter chord, where flow is believed to be attached for both cases. However the rate of deceleration of radial inflow in the control case decreased over the observed distance from the blade surface. In contrast, the obstructed case had the same deceleration at 5 cm and 10 cm away from the surface. The results show that the inboard-directed radial component of the inflow velocity decreases as the blade passes through the obstruction. This suggests that the centrifugal effect dominates, downstream of quarter chord. After the blade has passed through, the radial velocity returns to the background level.

7. Axial Flow and Cells of Radial Separation Initial measurements of axial flow showed that stall was indeed occurring, since the axial flow induced by a lifting surface was not occurring as the blade went through, or immediately thereafter. Further measurements resolve the mystery of why strong radial outflow does not continue to develop along the blade as was expected from the sharp decrease in radial inflow at the edge of the recirculation region. Instantaneous velocity fields show cells of spanwise recirculation separated by columnar flow structures, as schematically shown in Figure 12. It is important to distinguish the axial velocity directed away from the surface due to chordwise separation ( Figure 13) from axial flow due to the columnar flow structure and cells of spanwise separation. The extent of the flow structure is limited to only a couple of centimeters. The columnar flow is distinguished in contour plots of velocity by the presence of regions of opposite flow rotation on either side.

Figure 12. Columnar Flow in Stalled Blade

Figure 13. Example of Measurement of Recirculation

-0.5

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0.4

0 0.2 0.4 0.6 0.8 1

Chord, x/c

Ra

dia

l V

elo

cit

y U

R, m

/s

5 cm from upper surface

10 cm from upper surface

Figure 10. Radial Velocity Profile at Mid-Span without Obstruction (Control Case)

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

0 0.2 0.4 0.6 0.8 1

Chord, x/c

Ra

dia

l V

elo

city

U

R, m

/s

5 cm from upper surface

10 cm from upper surface

Figure 11. Radial Velocity Profile at Mid-Span with Obstruction

Figure 14. Instance of Recirculation at 0.25 x/c

Figure 15. Instance of Recirculation at 0.28 x/c

Figures 14 and 15 show two instances of the spanwise cells in the separated flow, as shown schematically in Figure 12. This process does not appear to be periodic at the rotor frequency, and thus differs substantially in location and strength from one rotor cycle to another. Thus this feature does not show up in averaged measurements. However, examination of several PIV frames confirms that the phenomenon persists.

8. Conclusions

A single-blade hover facility is used as a test-bed to develop hub-mounted instrumentation to study the temporal evolution of flow phenomena close to a rotating blade. An inflow obstructor plate is used to induce transient stall to simulate the dynamic stall phenomenon, given that the interest is in the region downstream of the separation line. 1. PLIS and PIV validate the experiment in that stall indeed occurs, based on the stoppage of the through-flow. 2. Downstream of the stall line, the radial velocity along the blade sharply develops an outward direction. However, the formation of stall cells along the radius serves to exchange fluid away from the blade boundary layer. 4. The phase-locked radial velocity profile shows higher deceleration of radial out-flow from leading edge to chord-wise location where flow recirculation begins. This suggests that while reattached flow’s inboard velocity component started to dominate over the centrifugal effect from leading edge to separation point, the inflow directional preference is nullified by the separation/recirculation zone created. 5. The cells of radial flow separation, while in the comparable scale as the blade chord, is time varying. This is possibly due to the dynamic nature of the extent of the radial out-flow due to domination of the centrifugal effect.

9. Acknowledgments The authors gratefully acknowledge support of this work through a grant from the Army Research Office. Dr. Thomas L. Doligalski is the Technical Monitor. Assistance in conducting the experiment by Martha Jaworski, Andrew Earl, undergraduate assistants, and in data reduction by Ms. Lan Wu, graduate research assistant, is acknowledged.

10. References 1 McCroskey, W. J., McAlister, K. W., Carr, L. W., and Pucci, S. L., “An Experimental Study of Dynamic Stall on Advanced Airfoil Sections,” NASA TM-84245, 1982. 2 Chandrasekhara, M.S., Wilder, M.C., Carr, L.W., “Compressible Dynamic Stall Control Using Dynamic Shape Adaptation”. AIAA Paper 99-0655, 1999. 3 McAlister, K. W. and Tung, C. “ Suppression of Dynamic Stall with a Leading-Edge Slat on a VR- 7 Airfoil”, NASA TP 3357, March 1993

4 Carr, L. W., “Progress in Analysis and Prediction of Dynamic Stall, Journal of Aircraft, Vol. 25, No. 1, January 1988, pp. 6 – 17. 5 W. J. McCroskey, W.J., “Some Current Research in Unsteady Fluid Dynamics - The 1976 Freeman Scholar Lecture” Journal of Fluids Engineering, Vol. 99, March 1977, pp. 8 - 38. 6 McCroskey, W.J., McAlister, K. W., Carr, L. W., Pucci, S. L. , Lambert, O. “An Experimental Study of Dynamic Stall on Advanced Airfoil Sections” Volume 1. Summary of Experiments, NASA TM 84245, July 1982. Vol. 2: Pressure and Force Data, September 1982, Vol. 3, Hot-Wire and Hot-Film Measurements, December 1982. 7 Van Dommelen, L. L. & Shen, S. F. (1982) The genesis of separation. In Symposium on Numerical and Physical Aspects of Aerodynamic Flows, (T. Cebeci, Ed.) 293-311. Springer-Verlag. 8 Lorber, P., McCormick, D., Anderson, T., Wake, B., McMartin, D., Pollack, M., Corke, T., Breuer, K., “Rotorcraft Retreating Blade Stall Control”. AIAA 2000-2475, Fluids 2000 Conference and Exhibit, Denver, CO. 9 Bousman, W., “A Quantitative Examination of Dynamic Stall from Flight Test Data”. Proc. AHS Forum, May 1997, p. 368-387. 10Bousman, W., “Airfoil Dynamic Stall and Rotorcraft Maneuverability”http://halfdome.arc.nasa.gov/publications /files/Bousman_TM.pdf. Also “Evaluation Of Airfoil Dynamic Stall Characteristics For Maneuverability” 26th European Rotorcraft Forum, The Hague, Sep. 26-29, 2000. 11 Coton, F. N., Wang, T., and Galbraith, R. A. McD., "An Examination of Key Aerodynamic Modelling Issues Raised by the NREL Blind Comparison ", Wind Energy, 5, 2002, pp. 199-212, ISSN 1095-4244 12 Corten, Gustave.P., “Flow Separation on Wind Turbine Blades”. Doctoral Dissertation, University of Utrecht, 2001, 152p. http://www.ecn.nl/docs/library/thesis/corten.pdf 13 Bunnes, P., Fiddes, S., “Laser Doppler Results Contributed to 'Dynamic Stall and Three Dimensional Effects', Joule II contractors meeting 8-10 feb 1995, NEL, Glasgow, U.K 14 Xu, Guanpeng, “Computational Studies of Horizontal Axis Wind Turbines”. PhD Thesis, School of Aerospace Engineering, Georgia Institute of Technology, May 2001. http://www.ae.gatech.edu/~lsankar/NREL/xu-diss.final.doc 15 Komerath, N.M., Liou, S-G., Hyun, J-S., "Flowfield of a Swept Blade Tip at High Pitch Angles". AIAA 91-0704, Aerospace Sciences Meeting, Jan. '91.