aiaa 2000-2475 rotorcraft retreating blade stall controlmacmardg/pubs/aiaa2000-2475.pdf · aiaa-...

AIAA- 2000-2475

0American Institute of Aeronautics and Astronautics

AIAA 2000-2475

Rotorcraft Retreating Blade Stall Control

P. Lorber, D. McCormick, T. Anderson, B. Wake, and D. MacMartinUnited Technologies Research CenterEast Hartford CT

M. PollackSikorsky Aircraft CorporationStratford, CT

T. CorkeUniversity of Notre DameNotre Dame, IN

K. BreuerBrown UniversityProvidence, RI

FLUIDS 2000 Conference and Exhibit19-22 June 2000Denver Colorado

AIAA- 2000-2475


ROTORCRAFT RETREATING BLADE STALL CONTROL

Peter F. Lorber, Duane C. McCormick, Torger J. Anderson,Brian E. Wake, and Douglas G. MacMartin

United Technologies Research CenterEast Hartford CT

Michael J. PollackSikorsky Aircraft Corporation

Stratford, CT

Thomas C. CorkeUniversity of Notre Dame

Notre Dame, IN

Kenneth BreuerBrown University

Providence, RI

Abstract*

Flow control to avoid or delay rotorcraftretreating blade stall can be an enabling technologyfor future high performance rotorcraft. Aerodynamicexperiments and computations have indicated thatappropriate unsteady excitation can delay boundarylayer separation and stall on airfoils. Work is inprogress to determine the control requirements forhelicopter rotor blades at full scale Mach numbers,Reynolds numbers, and with unsteady pitchingmotions. Compact, powerful, and efficient flowactuation and control systems will be needed. Threeactuation concepts were favorably evaluated duringinitial studies: electromechanical directed syntheticjets (DSJ), periodic flow modulation, and plasmaactuation. Electromechanical DSJ and plasmaactuators are being developed further and will beevaluated in full scale pitching blade sectionexperiments. These experiments will determine therequired control authority, validate the actuatorconcepts, and study open and closed loop controlapproaches. Computational studies are beingperformed of the combined external and actuator flowfields to determine preferred actuation geometries andoperating points. System analyses are being used toquantify the benefits for representative aircraftconfigurations and missions.

Copyright ÿ2000 by United TechnologiesCorporation. Published by the American Institute ofAeronautics and Astronautics, Inc., withPermission.

Nomenclature

A airfoil pitch rate, Uc 2/α&c airfoil chord

CL section lift coefficient, 25.0/ cUL ρCµ momentum coefficient,( ) ( )∞22 cUhu JET ρρF+ jet frequency, UcJETf

h jet slot widthk reduced frequency, Uc 2/ωM freestream Mach numberRe Reynolds number, ν/UcuJET jet unsteady velocity amplitudeU freestream velocity

IntroductionHigh performance future rotorcraft will require

significant improvements in multiple attributes,including increased range for global self-deployability, increased speed and performance forfast, agile missions, increased payload, and reducedexternal noise emissions, cabin noise and vibration.Analyses such as the Rotary Wing VehicleTechnology Development Approach (TDA) and theArmy After Next (AAN) have defined specificobjectives for military vehicles. One set of relevantobjectives includes:

• 24% increase in blade loading• 20% reduction in drag• 10% increase in aerodynamic efficiency• 112% increase in maneuverability/agility• 60% reduction in vibratory loads


Substantial improvements such as these will alsobe required to expand the use of civilian rotorcraftbeyond current applications. Since traditional passivedesigns appear incapable of producing sufficientlylarge simultaneous improvements in these attributes,active systems for controlling rotor loads,performance, and acoustics are being extensivelystudied. (Refs. 1-5).

Active flow control technologies can play asignificant role by providing the means to avoid ordelay retreating blade stall. Retreating Blade Stall(RBS) establishes limits on rotor load and flight speed.In addition to the loss of capability to generate lift,unsteady blade stall transmits very large impulsiveblade pitching moments to the flight control system.In order to prevent excess control loads, a stallboundary must be set to define the maximum bladeload capability as a function of rotor load and flightspeed. These limits impact maneuverability andagility as well as speed and payload.

The fluid mechanism involved in blade stall isboundary layer separation near the leading edge of therotor blade during rapid motion to high blade angle ofattack. This dynamic stall phenomenon has beenextensively studied using pitching airfoil and threedimensional blade sections at both model and fullscale (Refs. 6-8, Fig. 1) and by computations (Ref. 9-11). At full scale, reduced frequencies are typically k~0.05 to 0.15, with maximum instantaneous pitchrates of A~0.002 to 0.02. The relative external flowMach numbers are M = 0.3 -0.5, while the peak localMach numbers near the separation location of x/c=0.05 to 0.15 are typically M = 1.1 to 1.3. Full scaleReynolds numbers are Re~4x106, with transition toturbulence occurring ahead of separation.

Stalled flow locations on the rotor disc have beenidentified by surface pressure measurements duringflight test of a full scale aircraft (Ref. 12, Fig. 2), andby surface pressure, heat transfer, and tuftmeasurements during wind tunnel test of model rotors(Ref. 13, Fig. 3). Figure 4 is based on model rotorexperimental data for a level flight condition, andshows the relatively small stall regions for a moderateload condition, and significantly larger stall regions ata high load condition that is very close to the stallboundary. Figure 5 show similar stall patternsobtained from blade pressure measurements on amaneuvering aircraft (Ref. 12). The combination ofdetailed unsteady airfoil experiments andcomputations and global rotor measurements hasdefined the RBS phenomenon that must be controlledand the conditions under which the control is required.

A practical way of avoiding or significantlydelaying RBS has not yet been demonstrated under

flight conditions. Separation control must mitigate theeffects of the dynamic stall vorticity on blade pitching

Figure 1. Full scale pitching 3D rotor blade modelused for dynamic stall investigations (Ref. 8).

Figure 3. Pressure instrumented model rotor usedfor stall boundary investigations (Ref 13.).

Figure 2. Black Hawk aircraft used for pressureinstrumented rotor measurements of stall (Ref.12).


moment, and maintain or increase the available lift.This must be accomplished in an unsteady,compressible, high Reynolds number flow. The stallcontrol approach that is applied on the retreating bladeside must not create a significant drag penalty on theadvancing side, where the relative external Machnumbers are 0.6 to 0.85. Further, the approach mustbe practically incorporated into a rotating blade.

Mechanical high lift devices such as leading edgeslats have been investigated and found to be able todelay dynamic stall (Refs. 14-15), but publisheddesigns have tended to either create excess drag if leftin place on the advancing side or to be impractical todeploy and retract each rotor revolution. Steady flowcontrol methods such as air injection (Ref. 16) orsuction (Ref. 17) have also been investigated, buttransferring sufficient fluid from the fixed to rotatingsystems at acceptable levels of power and complexity

presents significant difficulties. Successful applicationof flow control to the rotor blade will require thedevelopment of new and innovative technologies thatintegrate a thorough understanding of the underlyingfluid dynamics with a robust and efficient actuation,sensing, and control system.

This paper presents results from a study of flowcontrol alternatives for RBS control, identifies someof the expected aircraft system benefits, and describesthe approaches currently being pursued to develop andvalidate practical RBS control as part of the DefenseAdvanced Research Projects Agency (DARPA)MicroAdaptive Flow Control (MAFC) program.

Flow Control TechnologiesUnsteady excitation has frequently been

investigated as a way to delay or avoid separation(Ref. 18). An extensive series of recent experimentswere conducted by Wygnanski and his current andformer students (Refs. 19-21). While the precisemechanism is not fully understood, the conceptinvolves low level periodic forcing to modulate theformation of vortices in a separating flow. At or nearan optimum frequency of F+~1, high streamwisemomentum flow is driven towards the surface,energizing the boundary layer and avoiding massiveseparation (Fig. 6). The periodic excitation has beenprovided by moving a mechanical element, byinjecting an unsteady jet of air, or by alternate suctionand blowing. For the later two methods, the keyparameter is the momentum coefficient, Cµ. While asubstantial quantity of work has been presented,definitive results for rotor blades at full scale flightconditions are not yet available.

The primary separation control technique beingapplied in the current work is the Directed SyntheticJet (DSJ). It applies acoustic streaming (Ref. 22) toform a synthetic jet (Ref. 23) with an exit neckoptimized for separation control (Ref. 24). Figure 7shows a typical DSJ configuration. The curved neckallows low momentum fluid to be ingested during thesuction phase of the DSJ and high momentum fluid tobe ejected during the blowing phase. Both phasesenergize the boundary layer. At high enough Cµ, the

U

Figure 4. Measured blade stall regions frommodel rotor for level flight at increasing rotorthrust (Ref. 13).

Figure 5. Stall regions identified from maneuveringaircraft blade pressure measurements (Ref. 12). Low level periodic forcing

modulates the vortex rollup At optimum F+, highmomentum flow is driventoward the wall, energizingthe boundary layer

Figure 6. Periodic excitation concept forseparation control.


DSJ can suppress separation without the need tooperate at an optimum F+. The concept and initialvalidation of the DSJ using a planar diffuser was morefully described in Ref. 24.

Initial Feasibility DemonstrationsA preliminary effort to evaluate dynamic forcing

at full-scale blade conditions was made at UTRC in1996, as part of an internally-funded effort toinvestigate active rotor and flow control techniques. Anear full scale blade section was modified to includean air supply and high frequency solenoid valves. Thepulsed air was expelled from a series of slots near 6%of chord. This technique succeeded in delaying

separation by 3° in steady flow at M=0.2 andRe~2x106, but was less effective at higher Machnumbers or during unsteady blade pitching motions. Itis believed that this was caused by insufficientunsteady momentum coefficients being produced bythe solenoid valve system (Cµ~. 0001). This initialsystem was severely constrained by the less than0.7in2 cross sectional area available ahead of the bladespar in this model.

The initial application of the DSJ to control airfoilseparation was performed in 1998 using a similarairfoil model in a low speed steady flow facility. Thelimited authority of the speaker-based actuationlimited this experiment to M~0.1 and Re~5x105.Figure 8 shows a series of smoke flow visualizationsat constant angle of attack (α=24°) and increasing Cµ.At moderate Cµ (0.002 to 0.005) the leading regionreattaches, at higher Cµ (0.01 to 0.025) the boundarylayer becomes essentially attached, and at extremelyhigh Cµ (0.04) the boundary layer is overdriven toform a wall jet. This experiment demonstrated howthe DSJ could operate as an unsteady excitation deviceat lower Cµ and as a separation suppression device athigher Cµ.

The Cµ level required for full scale rotor bladesremains uncertain. No published experiment has fullyduplicated the combined parameters of M, Re,unsteady motion, and airfoil contour. Combining the

Time-AveragedMass and Momentum

Flow

WithControl

Acoustically drivencavity

low x-mom high x-mom

Figure 7. Directed synthetic jet separation controlconcept

Leadingedge

Laser sheetintersectionwith airfoil

Flow

Reflectionof smoke

Side plateSideplate

Cµ=0.0025 Cµ=0.005

Cµ=0.010

Trailing edge

Cµ=0.040

Cµ=0

Cµ=0.015

Figure 8. Smoke visualization of DSJ effect on steady airfoil flowsα=24° (without blockage correction) andRe=1.8x105 (Ref. 24).


results such as Refs. 20, 21, and 24 implies thatCµ~0.0001 is ineffective, Cµ~0.001-0.002 may beacceptable, and Cµ~0.005-0.01 should work well.The approach that has been established for the currentprogram are to increase the blade section stall angle byat least 5° and increase CLmax by at least 10% usingCµ~0.001 to 0.002.

Actuation OptionsActuation is the most difficult technical barrier

that must be overcome to implement blade stallcontrol for a full scale helicopter. The followingparameters were specified to establish requirementsfor a RBS control system.

• 20 ft radius, 1.5 ft chord main rotor blades• Retreating blade relative Mach ~ 0.3 to 0.4• Apply control from ~40 to 85% of radius• Add less than 10% to blade mass• Must not compromise structural integrity• Power requirements < 1% of available power

This generated the following actuator requirements:

• F+~1 implies f ~300Hz• Cµ~0.001 implies uJET~200 fps

An actuator screening was performed in phase 1.Candidates included piezoelectric ceramics, fluidics,electromechanical synthetic jet, passive vibration-driven synthetic jet, periodic flow modulation, andplasma actuation, and microfabricated synthetic jetsand plasma actuators. Three appeared most promisingfor rotorcraft separation control: electromechanicalsynthetic jet, periodic flow modulation, and plasmaactuation. They were evaluated in detail and arediscussed below.

Electromechanical Synthetic JetsThe electromechanical devices considered here

are restricted to linear motors driven by anelectromagnetic interaction. Initially, three motorswere considered: moving coil, variable air-gap, andvariable reluctance. The moving coil uses the sameprincipal as audio loudspeakers: the force is controlledby an electromagnet coil moving within a stationarypermanent magnetic field. In the variable air gap andreluctance motors, the electromagnet coil is stationary.The coil generates a magnetic circuit through anarrangement of ferrous metal wherein part of thearrangement (the armature) is allowed to move. In thevariable air gap, the magnetic flux and force are inalignment creating an extremely strong and nonlinearforce at small gaps. In the variable reluctance motorthe magnetic flux and force motion are perpendicular,yielding a weaker, but linear force. Dynamic modelsof each type of motor were constructed to assist in the

evaluation. The main advantages of the variable airgap and reluctance motors were in reliability (nooscillating coil leads) and increased power capacity(coil heat generation can be more readily conductedaway). These benefits were out-weighed by themoving coil’s lighter moving mass, more manageabletolerances, and commercial availability.

To develop practical actuator embodiments, anelectro-acoustic model was developed for a DSJdriven by a moving coil. To validate and tune themodel, a prototype DSJ actuator was built using acommercially available motor with 2 inch coildiameter (Fig. 9). The photograph shows theassembled unit and some of the individualcomponents. The schematic drawing shows the

Flexure

Field assembly

Coil

Piston

Flexure

Coil & field assembly

RL

CV

CVB

MN RN

Figure 9. Initial prototype DSJ actuator for benchtest and model validation

UP

ZACT

RL

MN

RN

CV

UN

CVB

PAmb

2CSR

BLV

D

Figure 10. Electro-acoustic model.


assembly and identifies the components associatedwith the electro-acoustic lumped parameter model ofthe system shown in Fig. 10.

The “voltage” or potential driving the circuit inFig. 10 is the acoustic pressure and the “current” isvolume velocity equal to the piston volume velocity(UP). The voltage across the coil, V, has been groupedwith motor parameters (magnetic flux, coil length andresistance, and effective area) to form a pressure term.The parameters with subscript “N” refer to neck orslot terms (volume velocity UN=uJETAN, mass MN, andresistance, RN). The capacitor elements (CV, CVB) inthe circuit refer to volumes which act as an acousticcompliance. The circuit model is similar to thatdescribed in Ref. 24, with the addition of the acousticcompliance of the sealed backed cavity, CVB, and theleakage across the piston, RL. The model assumes theacoustic mass of the leakage to be negligible.

As given in detail in Ref. 24, the actuatorimpedance consists of a 2nd order system that containsmoving mass (MAT: moving mass of coil, piston,nearby air, and flexure), compliance (CAS: flexure),and resistance (RAS: e.g. structural, air viscosity, andeddy current losses). This system can be more simplyrepresented by a resonant frequency,

ATASMC10 =ω

and mechanical quality factor, QM = RAS ωo/MAT. Asdiscussed below, a high quality factor is critical forachieving practical actuation.

Swept-sine tests of the coil electrical impedancewith various degrees of acoustic loading were used toimprove model parameter estimates and validate themodel. For example, a “free-air” test without the slotcover and vented back eliminates all the elements inthe circuit model with the exception of ZACT. With thismeasurement, the mechanical quality factor of theactuator can be adjusted in the model to fit the data.(Alternatively, a hammer test with an open coil givesthe same results). Sequentially adding the slot coverand sealing the back cavity allows tuning of the neckloss coefficients and effective acoustic neck length.Variation of the drive amplitude is needed to adjustthe non-linear jet dump coefficients of the neck andpiston leakage. The prototype actuator was alsoinstrumental in identifying and mitigating unexpectedloss mechanisms such as eddy currents in the coilformer tube.

The efficiency, defined as the acoustic powerconverted to useful fluid power (UN

2RN) divided bythe actuator input power, is the primary metric beingmaximized within the installation constraints.Exercising the model over a range of parameters hasbeen useful for determining several design criteria.For example, aligning the design frequency to the

mass-spring actuator resonance yields the highestefficiency. Matching the Helmholtz mode(neck/cavity) with the mass-spring mode widens thepeak bandwidth but does not improve maximumefficiency, and results in an impractical cavity volume.Piston leakage, particularly when the back cavity issealed, results in large loss in efficiency. Minimizingmoving mass while maximizing mechanical qualityfactor and motor constant (BL/RC

1/2), provides thehighest efficiency. It is also very important for agiven actuator and slot to size the piston correctly inorder to provide the best impedance match betweenthe coil and the acoustic load (discussed below).

Once a moving coil motor has been selected forthe application, based on the maximum motor constantthat fits within space constraints and has minimumcoil mass, a criterion is needed to maximize the poweroutput. This criterion is illustrated in Figure 11. Themoving coil design point is at the intersection of theforce limit (dictated by the thermal dissipation limit soas not to melt the coil) and stroke limit (set by thelength of coil winding) where the power (product ofcoil force and coil velocity) is maximized. Sincesome of the coil force is used to overcome mechanicaland neck losses, the actual DSJ design point is belowthis point (at the stroke limit for reasonable losses).Since the neck velocity is specified by the design Cµ,the piston area is given by uNAN/uC. Hence, theavailable force given at the DSJ design point in Fig.11 determines the length of slot which can beoptimally driven by the motor. This process isessentially impedance matching the motor to theacoustic load.

To illustrate the importance of mechanical qualityfactor for practical full scale frequencies and movingmass, the above optimization process was evaluatedfor a BEI-LA13-12-000A (14 gram coil), moving a 10gram piston, operating at 225 Hz, slot width of 2.5mm, and additional jet dump neck losses of 50%

Force constraint(thermal dissipation limit)

Str

oke

limit

0 1 2 3 4 50

2

4

6

8

10

Coil velocity UC (m/s)

Fo

rce

f(N

)

Moving CoilDesign Point

DSJ DesignPoint

Mechanical andNeck losses

Figure 11. Moving coil actuator optimization


(KD=1.5) over a range of mechanical quality factors.The results are shown in Fig. 12 in term of fractions ofpower consumption versus QM. The figure shows thatbelow QM =15, mechanical and coil losses dominate,limiting efficiency to 20% or less. In order to achieveefficiencies above 30%, a mechanical quality factor of50 or more is needed. This is equivalent to a systemwith 1% or lower damping.

Figure 13 shows a preliminary design for a full

scale RBS actuator for wind tunnel testing, It uses aBEI-LA13-12-000A motor designed for a frequencyof 225 Hz and jet velocity amplitude of 218 f/s (F+ =1and Cµ = 0.001 for the 2 foot chord wind tunnelmodel). The motors are grouped in packs of threedriving one rectangular piston. The actuator moduleshave sealed back cavities to take advantage of theadded stiffness to reduce the required flexure stiffness.Each module is sealed from the neighboring moduleson the front, so that both in-phase actuation andactuation with a spanwise alternation of phase can beapplied. If alternating phase actuation provideseffective stall control, it would enable the use of otheractuation arrangements such as rotary motors.

Periodic Flow ModulationAn alternative to the synthetic jet for generating

unsteady Cµ is modulating net blowing produced byan air source. The modulation actuator can berelatively simple, providing increased reliability atreduced weight and complexity. The use of a steadyair supply modulated by a valve in the rotor blademakes a very simple and compact design. Aconceptual rotary valve design (Fig. 14) wasdeveloped based on two concentric cylinders withslots; the inner cylinder rotating to align these slots forflow out of the valve.

While the number of slots in the rotating cylindercontrols the RPM required to obtain a givenfrequency, the size of the slots controls the airflow andpressure drop. A nominal design compatible with therotor blade internal dimensions and Cµ flowrequirements has four slots in the rotating elementwhich spins at 4500 rpm to deliver 0.21 lb/hr/blade-ftof air at 300 Hz. This requires an inlet pressure of 2psid to meet the flow requirements. The powerrequirement to overcome friction was estimated to beon the order of 0.005 W suggesting that the valve

Mechanical Quality Factor, QM

10 20 30 40 50 60 70 80 90 1000

0.2

0.4

0.6

0.8

1.0

Fra

ctio

nof

tota

lpow

er

Delivered power

Coil losses

Mechanical Dissipation

Neck Losses (kD=1.5)

Figure 12. Effect of damping on actuatorefficiency

6”

Actuator ModuleFigure 13. Preliminary design of electromechanicalactuator for wind tunnel model.

~ 1 in.

ValveSeal (4)

Valvebody

Rotatingvalve spool(4 slots)

Fixedvalve cage(2 slots)

Blowingslot

Rotorleadingedgeassembly

Hollowspar

Figure 14. Implementation of a cylindrical rotatingvalve for periodic bleed air modulation.


could be operated by the equivalent of a commonhand-held air motor and powered by the air used forthe blowing process. Furthermore, this level of powerwas much less than the flow power and could beneglected in that analysis.

Engine bleed air was used as the air source in thisanalysis. It is readily available with no additionalequipment required other than plumbing to deliver thecompressed air to the rotor blade. However, with only2 psid required at the rotating cylinder, most of thecompression work that has been done by the engine onbleed air (at a nominal 59 psig) is lost in anirreversible expansion. Use of an ejector reduces thebleed airflow requirements and recovers much of thebleed momentum by entraining ambient airflow.However, this process increases the overall efficiencyby only about 50%. It should be noted that the use ofa low-pressure air source to overcome this problemintroduces difficulties of its own; in addition to theadded hardware, the large ducting required to deliverthe air would be impractical in this application.

Since flow modulation produces only positivevelocities, while a synthetic jet produces both positiveand negative velocities, higher flow power is required,as shown in Fig. 15. These losses, combined with theinefficiencies of using a high pressure air source resultin power requirements ~6-8 times greater than thoseof a synthetic jet to get the same level of flowmodulation for the retreating blade stall application.

Therefore periodic flow modulation was dropped fromfurther consideration for the RBS application.

Plasma ActuationPlasma actuation provides a third actuation path

that avoids the mechanical complexity ofelectromechanical or flow modulation actuators. Anextensive effort has been made on the development ofactuators which are based on the concept of producinga plasma in a localized region to generate unsteadyvortical disturbances. The plasma is produced overelectrodes located on a surface, and used to excite aboundary layer or separated shear layer. This type ofactuator has been used to excite instability modes inMach 3.5 laminar boundary layers (Refs. 25-26), andto excite shear layers in incompressible to transonicMach number jets. As part of the current MAFCprogram, plasma actuators were evaluated for the RBSapplication, and demonstrated the potential forproducing the required flow velocities. Theadvantages of the plasma actuators are (1) that theyare fully electronic, with no moving parts which canfatigue or limit the frequency band width, (2) theyhave a high energy density, and (3) they are fullyscalable in size, with MEMS scales providing someadvantages in their operation.

The plasma flow-actuator concept is illustrated inFig. 16. For this, an AC voltage is supplied to pair ofelectrodes. The electrodes are separated by a

V

t

0

0

Vs

Vpk

Vm

Vs

Vpk

Vm

)tsin(VV pk ω⋅=

( ))tsin(1VV pk ω+⋅=

pks V707.0V ⋅=

pks V707.0V ⋅=

Power for givenVpk-Vm

Power for givenVs

3pkVg/rA21.0P ⋅⋅=

3pkVg/rA25.1P ⋅⋅= 3

pVg/rA54.3P ⋅⋅=

3pVg/rA60.0P ⋅⋅=

Periodic blowing

Synthetic jet

Periodic blowing:

Synthetic jet:

Figure 15. The higher velocity peak associatedwith periodic blowing requires 6 times the powerconsumption to generate the same level of flowmodulation. Figure 16. Plasma flow actuator concept.


dielectric-insulator material (Ref. 27). In our firstapplication (Ref. 25), an air gap was used between theelectrodes. Presently, we use materials which havebetter properties than air. When the AC potentialacross the electrodes exceeds a threshold value(inversely proportional to the electrode spacing and afunction of the dielectric properties) the air forms aplasma in the region of highest current flux gradients.The plasma produces a body force on the ambient air,

2EFB ∇∝ . This amounts to a pressure gradient

region which induces a flow towards the plasma. Theno-penetration boundary condition at the wall resultsin a secondary flow away from the wall, which isrepresented in the bottom part of Fig. 16. Withelectrodes which are aligned with the flow direction,the secondary motion would result in counter-rotatingstreamwise vortices.

The magnitude of a mean flow componentproduced by plasma actuators, or the magnitude ofunsteady disturbances, can be greatly enhanced byphase-shifting the AC input to an array of neighborelectrodes. This has been done successfully,demonstrating the ability of these actuators to producehigh amplitude, alternating velocities at the wall. Weenvision plasma actuators to be used as a "driver" forsynthetic tangential jets in the separation controlapplication. This actuation technology is beingdeveloped further and will be applied during thevalidation experiment described below.

One of the features of plasma actuators that couldlimit performance is the high voltages required tooperate the device. This arises due to the need todevelop high electric field strength across thedielectric. One means to lower this voltage is to bringthe anode and cathode closer together by using thinnerdielectric layers. In an attempt to explore thispossibility, a plasma actuator has been fabricatedusing MEMS technologies (Ref. 28). The device,shown in Fig. 17 is comprised of two series ofinterdigitated gold electrodes separated by a thin (1micron) layer of polyamide (Kapton). The device wasfabricated, for convenience, on a silicon substrate,although any substrate might be used. The polyamidefilm serves to dielectrically isolate the two electrodeswhich are then excited with a high frequency ACsignal to create the glow discharge, as describedearlier. The device, as shown here, is purely forexploratory purposes and was not tested for any flowcontrol capabilities. It does, however, have someattractive features, primarily that the voltage requiredto achieve breakdown of the air is somewhat lowerthan in the "conventional" devices described above.Coupled to this is the need to modulate the drivingvoltage at a higher frequency than the O(1kHz)

required in the conventional devices. Preliminarytests of the MEMS plasma actuator indicate that theydo generate a plasma (visible to the eye) and that theapproximate scaling of voltage and frequency is aspredicted. However, more detailed experiments needto be conducted before a complete understanding ofthe device operation and potential benefits is achieved.

Airfoil-Actuator ComputationsUnsteady Navier Stokes simulations are being

used to evaluate separation control performance at fullscale conditions and aid in the design of the actuationsystem. An initial series of computations have beenmade for a ramping airfoil with an internal DSJexiting on the upper surface. The airfoil used in thesecalculations was the Sikorsky SC2110 airfoil, amodern design which emphasizes improved stallcharacteristics. The CFL3D Navier-Stokes analysiswas used. CFL3D had been used previously at UTRCto predict steady stall enhancement using steadyblowing. The Baldwin-Barth turbulence model wasused, since it has provided good agreement withexperiment in the prediction of the stall onset on fullscale blade sections. In addition to the external flow,the flows inside the slot and the plenum weremodeled. The objective is to improve the effectivenessof the slot and plenum geometries. The grid near thevicinity of the slot and plenum is shown in Fig. 18.The eight-block grid consisted of 40,083 grid points.An unsteady boundary condition was implemented atthe wall in the plenum that represents the DSJ piston.The velocity normal to the wall oscillated by aspecified amplitude and frequency to simulate theoscillating wall. The calculations shown here are atM=0.1,Re=900,000.

Figure 17. Optical micrograph and schematiccross section of MEMS fabricated plasmadischarge actuator.


At a low angle of attack, velocity magnitudecontours are shown in Figs. 19a and b during theoutstroke and instroke of the DSJ. During theinstroke, significant losses occur when the jet dumpsinto the plenum. The CFD analysis is being used toguide improvements in the internal geometry.

Figure 20 shows steady state computational andexperimental results for no blowing, andcomputational results for steady and unsteady jetblowing. The improvement in peak lift coefficient isdramatic for the unsteady jet case with Cµ = 1.5% andF+=1.0. For a ramping airfoil (constant rate increasein pitch at a rateA=0.005) the unsteady effects of thepitching motion provide a significant increase in thelift enhancement relative to steady flow. The unsteadylift coefficient histories versus angle of attack areshown in Fig. 21. This figure includes the no blowingcase and unsteady blowing cases with the DSJ exit attwo different locations: x/c = 0.015 and 0.064. Forthese calculations, Cµ=2%. The peak lift coefficienthas a modest increase of 20% with the DSJ located at

Figure 20.CL versus angle of attack for steadySC2110 airfoil. Comparison of steady andunsteady blowing.

Figure 18. Computational grid in the vicinity ofthe DSJ slot and plenum.

Figure 21. UnsteadyCL for ramping SC2110 airfoil,effect of DSJ location forCµ=2%.

Figure 19. Velocity magnitude contours during theout-stroke(a) and in-stroke(b) of the DSJ

b) DSJ Instroke

a) DSJ Outstroke


6.4% chord. Even though the gain in CLmax is nothuge, the stall is still significantly softer, which shouldimply a significant increase in the blade stallboundary.

Analysis of variations in the slot/plenumgeometry, the blowing frequency and the slot locationare underway to determine the most effective design.These calculations are being done at morerepresentative full scale conditions. It is important tonote that the SC2110 airfoil was designed for stall,and thus appears better behaved than the SSC-A09airfoil used in Refs 7,8, and 11. The stall is softer forthe SC2110 than it is for the SSC-A09, which isthinner and has a sharper leading edge. This may limitthe gains available from the DSJ. However, use ofthicker airfoils such as the SC2110 compromises thehigh Mach number performance. It may therefore bedesirable to apply the DSJ to a thinner airfoil toachieve good stall behavior without compromising thehigher Mach number performance.

Controllers and SensorsTwo modes of operation will be considered:

open-loop control and closed-loop control. The closedloop mode has the potential of using less power anddelivering more performance than the open loop modeat the expense of requiring separation sensors andlogic that is more complex. This tradeoff will bestudied in order to obtain the best solution to theproblem.

Open loop control modeThe simplest control mode is to activate the

actuator array during the flight conditions in whichseparation is known to exist. This control moderequires a schedule that determines theactivation/deactivation times based on the flightcondition. The control architecture for this mode issimple since no sensors that detect flow separation arerequired. Despite its simplicity, there are somedisadvantages with this approach. In particular, off-line analysis (based on models or heuristics) isrequired to determine the flight conditions in whichseparation takes place. Any flight condition in whichthe prediction of separation is incorrect will yield asub-optimal system---either the array will be activatedwhen it is not necessary (a power loss) or it will notfunction when required (a performance loss).

Closed loop control modeA more elaborate control mode is to activate/de-

activate the actuator array using sensors that detectflow separation. Pressure or surface shear sensorswould be used to detect flow separation. Thecombination of hot film shear sensors and pressuresensors has been shown to provide good definition of

the development of unsteady transition and separation(Refs. 8-10). To implement this approach a suitableseparation precursor signal needs to be constructedfrom the sensor information. A control law (or logic)that decides the activation/de-activation times basedon the precursor signal will be determined. Potentially,this control mode would require less power and yieldbetter performance, than the open loop mode.However, the stability of the overall system is nolonger guaranteed. The issues to be addressed in thismode are sensor selection and location, the generationof suitable precursors, the control law, and thestability analysis of the configuration.

The control parameters that affect the overallsystem behavior include activation and deactivationtimes (or azimuth angles), forcing frequency, andforcing amplitude. These parameters can be variedon-line to continuously optimize the systemperformance. Peak-seeking algorithms have beendemonstrated for other applications (Ref. 29), and insome low speed proof of concept separation controlexperiments. These algorithms rely only onmeasurement of an overall performance variable, andslowly conduct a gradient search over parameter spacefor the best solution. For current algorithms, thealgorithm time constant must be slower than thedynamics (here around 200-300 Hz) to average outtheir effects, and faster than any maneuver timeconstants (here, less than 5 Hz) that affect theoptimum solution.

Stall sensingAlternatives for sensing flow separation and blade

stall include local pressure and shear (heat transfer),strain, and vibration. Strain and vibration sensors onthe blade or on the aircraft should be able to readilydetect stall onset, but the indications will likely occurtoo late, after significant dynamic stall. Pressure andshear sensors on the blade surface may be more ableto detect precursors to separation or mild stall eventsin time for the control system to be activated toprevent serious stall. Figure 22 shows typical surfacepressure traces at several chordwise (x/c) stationsduring dynamic stall at full scale (Ref. 7-8). Twopossible stall indicators are called out. The first is arelatively high frequency oscillation that is oftenobserved prior to separation for M>0.3, when there islocal supersonic flow near the leading edge. Whiletimely, this indicator has small amplitude and may behard to distinguish. The second indicator is the rapidincrease in the suction (negative) pressure coefficientthat occurs while the stall vorticity is collecting. Thisindicator would be easier to detect by comparing thelocal pressure to a predetermined function of local


Mach number (derived from radius, rotor RPM, andflight speed).

Surface shear or heat transfer sensors are also ableto indicate stall events. Figure 23 shows pressure andhot film gage results for several stall events on an

oscillating airfoil at increasing mean angle of attack.Both the shear and pressure sensors provide largeidentifiable changes at stall.

RBS Control Benefits AnalysisThe impact of RBS control on conventional rotor

sustained lift limitations is shown in Fig. 24 as rotorthrust versus forward airspeed for baseline andenhanced. The figure normalizes all rotor thrusts bythat of the baseline rotor at a minimum powerairspeed. Enhanced rotor performance from a bladeflow separation control system is incrementally shownfor two moderate levels of improvement. The firstincrement represents a 5 degree increase in the stallangle of attack. The second increment adds anadditional 10% increase in CLmax. At a constantairspeed, the enhanced design rotor thrust limit isincreased by about 12% and this remains relativelyunchanged across the spectrum of airspeeds. At aconstant thrust, the enhanced design rotor speed limitis increased by about 40 knots at high airspeed and bymore at low speed. In Fig. 24, the enhanced rotorsystem benefit is extended to hover. While rotor stalldoes not normally limit aircraft system hovercapability (where power/torque available usuallydoes), an RBS control system would provide distinctbenefits during recovery maneuvers following anengine failure in hover, where rotor limits do play animportant role. A multitude of rotorcraft benefits froman RBS control system could exist that would enhanceperformance, mission effectiveness, maneuverabilityand survivability.

Aircraft mission effectiveness is enhanced bygains in performance. Payload, range and enduranceimprovements will result from increases in rotor thrust

Figure 22. Dynamic stall pressure indicators. Curvesshow pressure time histories at several chordwisestations during a stall event on an a pitching airfoil

Suction pressurecoefficientsCp < Cp(M)

Precursor pressure orshear oscillations

0 0.2 0.4 0.6 0.8 1.0

0 0.2 0.4 0.6 0.8 1.0

0

-1

-2

-3

-4

-5

Time, τ

Pre

ssur

e

Time, τ

4.95

4.65

4.90

4.70

4.80

4.85

4.75

She

ar

Stall Vortex

Separation

IncreasingMean α

Figure 23. Pressure and shear indicators. Curvesshow time histories for several stall events on anairfoil at increasing mean angle of attack

Figure 24. Rotor stall boundary (maximum thrustvs. velocity) curves for baseline and two levels ofRBS Control.

140 160 1800.80

0.85

0.90

0.95

1.0

1.05

1.10

1.15

1.20

1.25

0 20 40 60 80 100 120

VELOCITY - KNOTS

NO

RM

ALI

ZE

DT

HR

US

T

Control to 95%R:5° Stall Delay +

10% CLmax Increase

Baseline(No Control)

Control to 85%R:5° Stall Delay


margin. Utility helicopters are typically used totransport payload (cargo and/or troops) from onelocation to another. A normalized H-60 payloadversus range chart is shown in Fig. 25 to illustrate thismission. A 23% increase in payload would resultfrom a 12% thrust margin increase, assumingoperating weight limitations are increasedaccordingly. Defense aircraft are often required toperform specific operations at a remote location(radius point) prior to its return flight. This scenario isdepicted in Fig. 26 in the form of H-60 on-stationendurance versus radius of action. Endurance andrange enhancements are made possible by using theadditional thrust margin to increase the fuel load attake off where weight limitations would otherwiserestrict fuel. Another mission effectiveness scenariowhere RBS control benefits could be profound is thatof deployment. Helicopter ferry missions requireauxiliary fuel installations for extended range. Theseconfigurations often result in high operating weightsrequiring maneuverability limits to be imposed. Flightaltitudes restrictions also result when high weightsencroach on rotor limits. Extended range missionswould benefit significantly from an RBS controlsystem, not only in the form of improved rotorefficiency and its associated range, but also fromreduced flight times made possible by higher flightspeeds. For the examples analyzed, rotor efficiencybenefits were neglected in the calculations at increasedweights. Weight and power requirements of an RBScontrol system were also not included for these cases.

Maneuverability will also significantly improveas a result of the increased sustained rotor thrustmargins shown in Fig. 24. Nap of the earth (NOE)flight is dominated by low speed masking maneuversand quick accelerations and decelerations.

Conversely, air-to air combat requires high speedcapability and the ability to perform evasivemaneuvers, such as pull-ups and high speeddecelerating turns. Contour flight requires steady turnrate capability at moderate and high speeds. An RBScontrol system that improves speed capability andprovides a 12% increase in sustained rotor thrust willlikely increase turn rates by as much as 30%, increasemaximum load factor by up to 0.5g and increase bankangle limitations by 30 degrees. Gains of thesemagnitudes will dramatically enhance combatsurvivability.

IntegrationIntegration issues are significant for RBS control.

Existing blades are fabricated with a hollow compositeor titanium structural spar and a light weighthoneycomb trailing edge pocket, surrounded by afiberglass skin. The region ahead of the spar containscounterweights, erosion protection layers, and, ifrequired, deicing systems. The stall control systemmust either be located in the limited area ahead of thespar or located inside the spar with a fluid ormechanical passageway through the spar. In eithercase, a significant effort will be needed to avoidcompromising spar structural integrity.

Rotor blade control systems must meet severespecifications including vibration, temperature,durability, weight, size, and power required. Theymust survive the high centrifugal loads (400 g at midspan), vibratory loads, and environmental conditionssuch as precipitation, particle erosion. Fixed torotating frame power and information transfer systemsmust be incorporated.

The primary cost and reliability drivers will be

Figure 25. Normalized utility mission payloadcapability increase created by RBS control system.

0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

0 0.2 0.4 0.6 0.8 1 1.2

NORMALIZED RANGE

NO

RM

ALI

ZE

DP

AY

LOA

D

TakeoffCruise At Best Range SpeedLand With Reserve

23% Increase In Payload

Figure 26. Normalized mission range andendurance enhancement created by RBS controlsystem.

NORMALIZED RADIUS

NO

RM

ALI

ZE

DT

IME

-ON

-ST

AT

ION

0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

Baseline:Available Fuel

Limited byGross Weight

35% Increase in Max Range andEndurance for 33% Added Fuel

TakeoffCruise at Max SpeedOn Station(Hover/Loiter/Cruise)Return CruiseLand with Reserve


added overall system complexity, which may increasedesign, fabrication, and maintenance requirements.

Large Scale Validation ExperimentThe most significant issues that must be

addressed in order to determine the feasibility ofapplying flow control to improve rotor blade retreatingblade stall characteristics are

1. what unsteady momentum coefficient amplitudesand frequencies are sufficient to achievesignificant improvements at realistic conditions?

2. can actuation systems be developed to providethis momentum coefficient within the size, massand power constraints of a rotor blade?

3. can this system be integrated within a rotor bladeand adapted to survive and be maintained in itscentrifugal load, vibratory load, and thermalenvironment?

As a first step, a large scale validation experimentis being developed to answer questions 1 and 2. Theexperiment will be performed on a full scale (M = 0.3to 0.4, Re= 3-5x106), two dimensional blade sectionmodel that can be oscillated at the appropriatefrequencies (f~4-6 Hz, k=0.05 to 0.15) and amplitudes(± 10°) at mean angle of 0 to 20 deg). Figure 27shows a drawing of the 24 inch chord, 33 inch spanblade section model. The airfoil section will be theSC2110, which is used on the inboard sections of newSikorsky rotor designs.

The experiment will be conducted in a twodimensional channel (TDC) mounted in the 8'octagonal test section of the UTRC Main WindTunnel, as shown in Fig. 28. This section is capableof maximum Mach numbers up to 0.75-0.8, providingthe capability of not only validating the separationcontrol system under retreating blade conditions, butalso of determining any impact on section drag underadvancing blade conditions (M=0.7 to 0.8).

Unsteady pressure sensors will determinechordwise pressure distributions, lift, pitchingmoment, and pressure drag coefficients. Surface shearsensors will determine boundary layer state. Bothsensors will be available for closed loop controlexperiments. A wake rake will measure section drag.

Two actuation approaches will be applied. Thefirst experiment, scheduled for Fall 2000, will useelectromechanical synthetic jets located inside themodel. This experiment is intended to validate theassumptions of required Cµ , frequency, and jet exitslot location, and determine how well the actuationsystem performs. The second experiment, scheduledfor mid 2001, will use both improved

electromechanical synthetic jet actuators and plasmaactuators. This experiment will demonstrate a moreflight representative actuation system, and evaluateopen and closed loop control architectures.

Concluding RemarksSignificant rotorcraft performance benefits can be

achieved by implementing retreating blade stallcontrol. Flow control techniques such as synthetic jetsand plasma actuation are being developed to effectsuch control at acceptable levels of mass, power, andcomplexity. Specific actuation, control, and sensing

Figure 27. Drawing of 2D wind tunnel modelblade section, showing DSJ slots.

Figure 28. Wind tunnel test section two dimensionalchannel (TDC) apparatus for full scale pitching bladesection validation experiment

8’ Octagonal Test Section

BladeSection

TwoDimensional

Channel

Blade PitchOscillationActuation


embodiments are currently being designed andevaluated. Initial experimental and computationaldemonstrations have yielded encouraging results.Validation experiments are planned at full scalehelicopter rotor blade conditions to demonstrateaerodynamic benefits and actuation and controlsystem performance.

AcknowledgementsInitial stages of this rotorcraft flow control effort

were supported by internal UTRC IR&D programs.Current work is supported by the DARPAMicroAdaptive Flow Control Program (WilliamScheuren and Richard Wlezian) under U.S. ArmyResearch Office (Thomas Doligalski) ContractDAAG55-98-C-0066.

References1. Brooks, T.F., and Booth, E.R., “The Effects of

Higher harmonic Control on Blade VortexInteraction Noise and Vibration”,Journal of theAmerican Helicopter Society,Vol.38, No.3, July1993.

2. Friedman, P.P. and Millot, T.A., ”VibrationReduction in Rotorcraft using Active Control- AComparison of Various Approaches”, AHSAeromechanics Specialists Conference, SanFrancisco, CA Jan. 1994.

3. Dadone, L, “An Overview of Benefits Possiblefrom the Application of Smart Structure materialsto Rotor Blades”, ARO 2nd Workshop on SmartStructures and Materials, U. Maryland, Sept.1995.

4. Jacklin, Blaas, Teves, and Kube, “Reduction ofHelicopter BVI Noise, Vibration, and PowerConsumption Through IBC”, AHS 51st AnnualForum, Ft. Worth TX, May 1995.

5. Cheng, R.P., Theodore, C.R., and Celi, R.,"Effects of Higher Harmonic Control on RotorPerformance", AHS 56th Annual Forum, VirginiaBeach VA, May 2000

6. Carr, L.W. and Chandrasekhara, M.S.,“Compressibility Effects on Dynamic Stall”,Progress in Aerospace Sciences, Vol. 32, pp. 523-573, 1996

7. Lorber, P.F., and Carta, F.O., “Airfoil DynamicStall at Constant Pitch Rate and High ReynoldsNumber,”Journal of Aircraft, Vol. 25, June 1988,pp. 548-556.

8. Lorber, P.F. “Compressibility Effects on theDynamic Stall of a Three-Dimensional Wing,AIAA Paper 92-0191, 30th Aerospace SciencesMeeting, Jan. 1992.

9. Visbal, M.R., "Effect of Compressibility onDynamic Stall of a Pitching Airfoil", AIAA paper88-0132, 26th Aerospace sciences Meeting, Reno,NV, Jan. 1988

10. Grohsmyer, S.P., Ekaterinaris, J., and Platzer,M.F., "Numerical Investigation of the Effect ofLeading Edge Geometry on Dynamic Stall ofAirfoils", AIAA paper 89-0024. 27th Aerospacesciences Meeting, Reno, NV, Jan. 1989

11. Patterson, M.T., and Lorber, P.F.,“ Computational and Experimental Studies ofCompressible Dynamic Stall,”Journal of Fluidsand Structures, Vol. 4, 1990, pp. 259-285

12. Bousman, W.G., "A Qualitative Examination ofDynamic Stall from Flight Test Data",Journal ofthe American Helicopter Society, Vol. 43,October 1998, pp. 279-295.

13. Lorber, P.F., Stauter, R.C., Haas, R.J., Anderson,T.J., Torok, M.J., and Kohlhepp, F.J.,“Techniques for Comprehensive Measurement ofModel Helicopter Rotor Aerodynamics, 50thAnnual Forum of the American HelicopterSociety, May 1994.

14. Carr, L.W., and McAlister, K, “The Effect of aLeading Edge Slat on the Dynamic Stall of anoscillating Airfoil”, AIAA Paper 83-2533,AIAA/AFS Aircraft Design, Systems, andOperations Meeting, Ft. Worth TX, Oct. 1983

15. Narramore, J.C., McCroskey, W.J., and Noonan,K.W., ”Design and Evaluation of Multi-ElementAirfoils for Rotorcraft”, AHS 55th Annual Forum,Montreal, Quebec, May 1999.

16. Yu, Y.H., McAlister, K.W., Tung, C., and Wang,C.M. “Dynamic Stall Control for AdvancedRotorcraft Application”, AIAA Journal, Vol.33,No. 2, Feb. 1995.

17. Karim, M.A. and Achyara, M., “Control of theDynamic Stall Vortex over a Pitching Airfoil byLeading Edge Suction”, AIAA Paper 93-3267,Shear Flow Conference, Orlando FL, July 1993.

18. Koga, D.J., Reisenthal, P., and Nagib, H.M.“Control of Separated Flows Using ForcedUnsteadiness” Illinois Institute of TechnologyFluids & Heat Transfer Report R84-1, 1984

19. Katz, Y., Nishri, B., and Wygnanski, I., “TheDelay of Turbulent Boundary Layer Separation


by Oscillatory Active Control,” AIAA Paper 89-0975, 1989

20. Seifert, A. and Pack, L.G., “Oscillatory Controlof Separation at High Reynolds Numbers”,AIAA Paper 98-0214, 36th Aerospace sciencesMeeting, Jan. 1998

21. Greenblatt, D., and Wygnanski, I., "ParametersAffecting Dynamic Stall Control by OscillatoryExcitation," AIAA Paper 99-3121, 17th AppliedAerodynamics Conference, Norfolk VA, 28 June-1 July 1999.

22. Ingard, U, “On the Theory and Design ofAcoustic Resonators”,Journal of the AcousticalSociety of America,Vol. 25, No. 6, November1953

23. Amitay, M., Smith, B.L., and Glezer, A,“Aerodynamic Flow Control Using Synthetic JetTechnology”, AIAA 98-0208, January 1998

24. McCormick, D.C., "Boundary Layer SeparationControl with Directed Synthetic Jets." AIAAPaper 2000-0519, 38th Aerospace SciencesMeeting, Reno, NV Jan 2000.

25. Cavalieri, D. "On the Experimental Design forInstability Analysis on a Cone at Mach 3.5 and 6Using a Corona Discharge Perturbation Method",M.S. Thesis, Illinois Institute of Technology,1995

26. Corke, T. C. and Cavalieri, D. "Controlledexperiments on instabilities and transition toturbulence in supersonic boundary layers", AIAAPaper 97-1817, 1997.

27. Massines, F. Rabehi, A., Decomps, P. Gadri, R.Segur, P. and Mayoux, C. "Experimental andtheoretical study of a glow discharge atatmospheric pressure controlled by a dielectricbarrier", J. App. Physics, Vol. 83, No. 6, pp.2950-2957, 1998.

28. Bayt, R,L. and Breuer, K.S. PrivateCommunication. 2000

29. Banaszuk, A., Zhang, Y, Jacobson, C.A.,“Adaptive Control of Combustion InstabilityUsing Extremum-Seeking”, 2000 AmericanControl Conference, June 2000.

aiaa 2000-2475 rotorcraft retreating blade stall controlmacmardg/pubs/aiaa2000-2475.pdf · aiaa-...

Documents