closed-loop control of separatedloop control of separated ...e9es6-7d... · aiaa june 2010 yaw\aiaa...

Closed-Loop Control of SeparatedClosed-Loop Control of Separated Turbulent Flows with Dynamically

Articulating Geometries presented atpresented at

GDR <<Flow Separation Control>> Meeting06-07 December 201006-07 December 2010

Futuroscope, Poitiers, France byby

R. Wallace, P. Shea, and M. GlauserV Thi kk d H C l

Syracuse University

Cle S ie e CV. Thirunavukkarasu and H. CarlsonRyan Schmit

Clear Science Corp.

Air Force Research Laboratory

F T t f A tiFocus on Turrets for Aero-optics ApplicationsApplications

AIAA June 2010 YAW\AIAA June 2010Updated_dec.ppt

APS 2010 Turret SARL\APS 2010.ppt

Shea_APS_2010.pptx

All material has been cleared for Public Presentation

Closed-loop Flow Control for an Articulating Turret with Two Degrees ofArticulating Turret with Two Degrees of

Freedom: Pitch and Yaw

R Wallace P Shea and M Glauser Syracuse UniversityR. Wallace, P. Shea, and M. GlauserV. Thirunavukkarasu and H. Carlson Clear Science Corp.

The Issuee ssueThe performance of lasers is degraded

th li ht th h t b l tas the light passes through turbulent regions such as wakes and shear layers

Wh f i t tWhy of interestOn board aircraft lasers housed within a 3D turret

• Aero-opticsK Gilbert 1982

Boeing website

– K. Gilbert, 1982• Aero-optic effects

– E. Jumper et. al. 2001• Adaptive optic control

– A. J. Smits and J. P. Dussauge, 2006• Strong velocity and density fluctuation correlationg y y

Previous Active Control Aero-Optic WorkPrevious Active Control Aero Optic Work

• B. Vukasinovic et al., 2009– Open Loop Control using

synthetic jets

Vukasinovic et al 2009Vukasinovic et al., 2009

• S. Gordeyev et al., 20052005

• Passive control

Gordeyev et. al., 2005

Previous Work: Closed Loop Control

Simple Proportional Closed Loop Control

Andino et. al 2008, Andino et. al 2009, Wallace et. al 2009, Andino et al, (to appear, AIAA Journal, January 2011)

Simple Proportional Closed Loop Control( )( )00

12sin)()( ttftaKtu

M

nn −⎥

⎦

⎤⎢⎣

⎡−= ∑

=

πClosed Loop Control input:

Flow driven towards homogeneity and integral time scales significantly reduced

Control Input: Synthetic Jets

scales significantly reduced

Motivation for Suction I ti ti b D t h d Edd Si l ti h dInvestigation by Detached Eddy Simulation showed suction gives an improvement in reducing the Optical Path Difference RMS and controlling the separationPath Difference RMS and controlling the separation bubble over the aperture

Reduction of OPDrms with suctionVorticity for baseline and a constant suction

Work done by Clear Science Corporation supported by a Phase I SBIR

Previous WorkClosed Loop PitchingClosed Loop Pitching

Simple Proportional Feedback Controller

Feedback Sensor Velocity

Plane

[ ]rmspKtu −=)(Actuator InputU∞

Plane

Flow Control OnNo Flow Control

Closed Loop Pitching with Advance ControllerController

Development and Implementation of a Reduce Order Model

Dynamical Estimator with Kalman filter Multiple pressure sensors

Velocity data

Controller gains calculated using a Linear Quadratic Regulator

Key Advantage of the Dynamical Estimator ControllerAble to keep the flow attached with a higher efficiency

T. Vaithianathan, et. al, “Feedback Flow Control for a PitchingTurret (Part I)," AIAA Paper 2010-0360, 48th AIAA Aerospace Sciences M ti O l d FL 2010

For further details please see:Meeting, Orlando, FL, 2010.

R. D. Wallace, et. al, “Feedback Flow Control for a PitchingTurret (Part II)," AIAA Paper 2010-0361, 48th AIAA Aerospace Sciences Meeting, Orlando, FL, 2010.

Objective

• Reduce the velocity fluctuations in the wake over the aperture of a dynamically azimuthal rotating turret using unsteady g g ysuction flow control

Active Flow Control– Active Flow Control• Open Loop

Cl d l t l• Closed loop control

Due to the strong velocity and density fluctuationsDue to the strong velocity and density fluctuations correlation, the strategy for reducing velocity fluctuations can be applied to reducing density fluctuationsfluctuations

Test Conditions• Facility

– Syracuse University y ysubsonic wind tunnel

• TurretTurret– Hemisphere with flat aperture– AR = 1.3– Quarter ScaleQ

• ActuationS ti l t– Suction slots

• Around the aperture• Double row

Suction Slots

• Flow Conditions– Mach number ≈ 0.1

Re ≈ 500 000 Suction Slots– Re ≈ 500,000

Control Inputp• Suction Actuators

U t d V l– Unsteady Valves• On/Off operation only• Duty Cycle modulation for y y

control– Relation between Duty

Cycle and Velocity

Cµ

• Operation speed of 25 Hz

– Vacuum SourceV t d b t• Vacated by two vacuum pumps

• Pressure 0.9 psi Mean Coefficient of Momentum per slot– Coefficient of Momentum

• CµMAX = 1.68x10-4 per slot

Mean Coefficient of Momentum per slot

Measurement B.L.Pressure

Transducer

Pressure Transducer

• Dynamic Surface U

Transducer

y a c Su acePressure– Sampled at 10,000 Hz

U∞

Sampled at 10,000 Hz– 30 acoustic ICP pressure

transducers– Located on and around aperture

Pressure Transducer Locations ObservationPurpose

OnlyLocated on and around aperture

• Velocity– Stereo Particle Image VelocimetryStereo Particle Image Velocimetry– Centerline Plane

• Pressure and Velocity essu e a d e oc ySimultaneously SampledSampled

Dynamic Yawing: Velocity No ControlVelocity, No Control

Yawing Turret

Azimuthal Range0° to 10°

Static Elevation AngleStatic Elevation Angle115°

⎤⎡⎟⎞

⎜⎛ω

⎥⎦

⎤⎢⎣

⎡⎟⎠⎞

⎜⎝⎛−= tπωθ cos5100

Dynamic Yawing: Pressure, No Control

Fluctuating surface static pressure time series

Open Loop Control: VelocityVelocity

A constant signal of 50% duty cycle at 25 Hz

Baseline Flow

No Control

Open Loop Control

Open Loop Controller Results: Pressurep p

No Control Open loop ControlNo Control Open loop Control

Asymmetry in the flowDue to the strong asymmetry seen in theDue to the strong asymmetry seen in the flow the suction slots are divided into two halves.

Examine the use of multiple outputs

Closed Loop Flow Control• Measurement Estimator

– Pod/mLSE Estimator• J. Pinier et al., 2007

• Simple Proportional Feedback Controller is employed

⎥⎦

⎤⎢⎣

⎡−= ∑

=

M

nn taKtu

1)()(&⎦⎣ n 1

is the POD/mLSE expansion coefficients)(tan

Only the first mode is utilizedy

Block Diagram

Multiple Input Multiple Output Controller

Right Sensors and valves

Left Sensors and valves

Feedback signal time series

Sensors off aperture were used

Closed Loop Pitching Results: Velocity

No Control Closed Loop Flow Control

Closed Loop Controller Results: PressurePressure

No Control Closed loop ControlNo Control Closed-loop Control

Closed Loop Yawing Results

( ) ( ) ( )[ ] ( ) ( )[ ]{ }⎟⎟⎠

⎞⎜⎜⎝

⎛−+−≡ ∑

=

PIVN

iiiii

PIVrms xutxuxutxu

Ntu

1

222

211 ,,1Spacial Velocity rms

DCuu

BaselinermsControlrms −=ξController Efficiency

DC

Conclusion

– Dynamic yawing produces a strong asymmetry over the turretasymmetry over the turret

– Open Loop control reduced velocity fluctuationsfluctuations

– Closed loop control had the a greater d ti f l it fl t ti threduction of velocity fluctuations than open

loop control

Current Closed Loop Control InvestigationsConducted at the Subsonic Aerodynamic Research Laboratory wind tunnel at Wright-Patterson Air Force BasePatterson Air Force Base

Freestream Velocity = Mach 0.3

Reynolds Number of 2,000,000

The Test ModelControl input of suction

Similar geometry as SU tests

Capable of two degrees of freedom: Pitch and Yaw

SARL Tests Goals

ExperimentsExperiments•Baseline

•Open Loop ControlOpen Loop Control

•Closed Loop Control

Train and implement a reduced order model Dynamical Estimator with a Kalman filter intro a feedback closed loop controller

Compare the advantages of the various control schemes

SARL MeasurementsVelocity/surface pressureVelocity/surface pressure

OPD/Surface Pressure

A k l d tAcknowledgementsSupport through a Phase II SBIRSupport through a Phase II SBIR

James Myatt and Ryan Schmit project mangers

Questions???

CHARACTERIZATION OF A THREE-DIMENSIONAL

TURRET WAKE FOR ACTIVE FLOW CONTROLTURRET WAKE FOR ACTIVE FLOW CONTROL

PART II: EXPERIMENTAL STUDY

Patrick R. SheaChris J. Ruscher     Ryan D. Wallace    

Mark N Glauser John F Dannenhoffer IIIMark N. Glauser     John F. Dannenhoffer, IIISyracuse University; Syracuse, NY

APS Division of Fluid Dynamics MeetingNovember, 2010

INTRODUCTIONI O U IO

Axisymmetric bluff bodies, typically referred to as turrets, are commonly used for optical housings on airborne platformscommonly used for optical housings on airborne platforms

A bl

cnet.com  (Photo credit: Ed Turner, Boeing)

Aero‐optics problem— When density fluctuations are present, laser performance on moving 

platforms is degraded as light passes through turbulent regions such as wakes and shear layers

PREVIOUS RESEARCHTurret flow fields

— LDA in the turret wakeA i e u e a eLeder et al. (2003)

— Force measurementsSnyder et al (2000) and Sluder et al (2008)Snyder et al. (2000) and Sluder et al. (2008)

— Surface flow visualizationGordeyev and Jumper (2010)

Aero‐optics active control research— Synthetic jet actuators

Gordeyev et al (2009) saw up to a 34% reduction in OPDRMSGordeyev et al. (2009) saw up to a 34% reduction in OPDRMSat Mach 0.3Andino et al. (2008) saw up to 19% reduction in surface pressure fluctuations on turret aperture at Mach 0.3p

— Dynamic suctionWallace et al. (2010) saw up to 48% reduction in mean RMS velocities above turret aperture at Mach 0.1velocities above turret aperture at Mach 0.1

EXPERIMENTAL TEST SETUPELow speed wind tunnel

Square test section of 0 61 m— Square test section of 0.61 m — Operated at 53 m/s

ReD ≈ 5 x 105

Turret— Axisymmetric geometry— Base diameter (D) of 0.152 m— Suction based active control— Fixed aperture angle of 120°Dynamic pressure sensors

— Sensitivity of 1500 mV/psiR l ti f 0 02 i— Resolution of 0.02 mpsi

— Resonant frequency ≥ 13 kHzDantec PIV system

Two 8 bit HiSense cameras U∞— Two 8‐bit, HiSense cameras— New Wave Gemini Nd:YAG laser— Double exposure/double frame 

cross‐correlation

∞

TURRET CONFIGURATION

Active Flow Control— Suction around leading portion of aperture— cμ≈ 3.9 x 10‐4 per slot

U∞

PIV CONFIGURATION

Two‐dimensional data taken at center plane of turret ll l t th f tparallel to the free‐stream

— Composite data from 8 inspection regions500 snapshot ensemble average for each region

LaserLaser

Laser SheetInspectionRegion

yU

Laser Sheet g

x

U∞

Pressure SensorLocations

STREAMWISE VELOCITY CONTOURS

Baseline

OpenLoop Control

STREAMWISE VELOCITY CONTOURS

Baseline

OpenLoop Control

AUTO‐SPECTRAL DENSITY FUNCTION

UU∞

B d & Pi l (1980)Bendat & Piersol (1980)

AUTO‐SPECTRAL DENSITY FUNCTION

UU∞

AUTO‐SPECTRAL DENSITY FUNCTION

UU∞

CONCLUDING REMARKS

PIV and dynamic pressure measurements have been k i h k f i h d i h itaken in the wake of a turret with and without active 

flow controlA ti e flo o t ol effe ti ely edu ed the i e of theActive flow control effectively reduced the size of the wakeDynamic pressure measurements indicated changes inDynamic pressure measurements indicated changes in the spectral characteristics of the wake flow field

— Changes in the spectral content were not consistent g pthroughout the wake

— Properly placed sensors can potentially be used for closed‐loop feedback flow controlloop, feedback flow control

ACKNOWLEDGEMENTS

AFOSR Grant with Clarkson UniversitySyracuse University Fellowship

Related upcoming talks from Syracuse University— Local Flow Control for Active Building Facades

Session EJ: Flow Control IIISession EJ: Flow Control III— A Closed‐loop Suction Flow Control Study over a Pitching Turret

Session EJ: Flow Control III