Rotorcraft Center of Excellence Analysis and Control of the Transient Aeroelastic Response of Rotors...

Post on 11-Jan-2016

219 views 1 download

Tags:

transcript

Rotorcraft Center of Excellence

Analysis and Control of the Transient Aeroelastic Response of Rotors During Shipboard Engagement and

Disengagement Operations

Jonathan A. KellerRotorcraft Fellow

Ph.D. Thesis SeminarMarch 22, 2001

• Introduction

• Previous Research

• Objectives

• Approach

• Results

• Conclusions

Presentation Outline

Introduction

•Unique challenges in ship-based operation of helicopters

­Small, moving deck area

-Strong & unsteady winds (often up to 50 knots)

-Unusual airflow patterns around ship decks

•Engagement (startup) of rotor system not mundane

Low RPM = Low CF

High winds = Potentially high Aerodynamic Forces

High blade flapping

(%NR)

Historical Motivation

•Past problems for Sea King (RN) and Sea Knight (USN)

-Blade-to-fuselage contact (114 for H-46!) - High blade loads

•Forces conservative limits to be placed on wind conditions

conducive to safe engagement operations

•Reduces operational flexibility of helicopter

H-3 Sea KingH-46 Sea Knight

Engage/Disengage Testing

• Safe conditions were determined in at-sea tests

­ Tests for every ship/helicopter/landing spot combo, but:

• Problems often occurred within “safe” envelopes

• Engage/Disengage testing cancelled in 1990

• Analytic methods needed!

Took 5 days, 15 people, $150k

No control of winds or seas

Calm weather = wasted tests

Styrofoam Pegs

GreasyBoard

Present Day Motivation

• H-46 tunnel strikes still frequently occur­ At least 3 last year aboard LHD type ships

• Use of Army helos on Navy ships (JSHIP Program)­ Army helos not designed for naval ops - no rotor brake?

­ Apache elastomeric damper loads during startup

­ Broken flap stop for Blackhawk during engagement op

­ Chinook is much like Sea Knight

H-47 ChinookH-46 Sea Knight

•USMC and USN (hopefully) purchasing V-22

­V-22 blades much shorter than articulated blades

»Excessive rotor gimbal tilt angles may be a possibility

»Contact with between blades & wing/fuselage not a concern

»Contact between gimbal and restraint potentially high loads

Rubber Spring

Gimbal Restraint

Future Motivation

Introduction

• Previous Research

• Objectives

• Approach

• Results

• Conclusions

Presentation Outline

Early Engage/Disengage Research

•Willmer, Burton, & King at Westland (1960s)

­ Investigated Whirlwind, Wasp, and Sea King helicopters

•Leone at Boeing Vertol (1964)

­ Investigated H-46 Sea Knight tunnel strikes

­ Measured and predicted loads during blade-droop stop impacts

•Healey et al at Naval Postgraduate School (1985-1992)

­ Measured model-scale ship airwake for LHA, DD, AOR

­ Unsuccessfully investigated H-46 Sea Knight tunnel strikes

•Kunz at McDonnell Douglas (1997)

­ Investigated high loads in AH-64 Apache elastomeric dampers

Recent Engage/Disengage Research

•Newman at University of Southampton (1985-1995)

­ Developed elastic F-T code for single rotor blade

» Articulated or hingeless hubs

­ Articulated rotors more prone to blade sailing than hingeless

­ Correlated code w/ model-scale rigid R/C helicopter tests

• Geyer, Keller, Kang and Smith at PSU (1995 - Present)

­ Developed F-L-T code for multiple rotor blades

» Articulated, hingeless, teetering, or gimballed hubs

­ Simulated H-46 Sea Knight engagements and disengagements

• Botasso and Bauchau (2000)

­ Multi-body modeling of engagement and disengagement ops

Introduction

Previous Research

• Objectives

• Approach

• Results

• Conclusions

Presentation Outline

Objectives

•Develop unique “in-house” analysis code to:

- Increase physical understanding of engage/disengage behavior

- Accurately predict safe rotor engage/disengage envelopes

Safe Region

- Control rotor response to expand engage/disengage envelopes

Unsafe Region

wtip (%R)

Technical Barriers

• Limited data of engage/disengage ops or ship airwake

• Simulation of a complex transient aeroelastic event

- Rotor speed is a function of time 0 and (t)

» Flap/lag stop or gimbal restraint impacts at low

- Complicated ship airwake and aero environment high ,,

ShipAirwake

H-46 Data

(%NR)

H-46 Data

Introduction

Previous Research

Objectives

• Approach

• Results

• Conclusions

Presentation Outline

Ship Airwake Modeling

• Specify speed (VWOD) & direction (WOD) relative to ship center

• Determines ship airwake (Vx, Vy, and Vz) in plane of rotor

»Vx, Vy in plane velocities, Vz vertical velocity

VWOD

Vz

Vy

Vx

WOD

Simple Ship Airwakes

• Simple airwake types derived from tests (Ref. Newman)

Vz = vertical velocity, = “gust” factor

LinearAirwake

VWOD

ConstantAirwake

HorizontalAirwake

VWOD

VWOD

Vz = VWOD

Vz

Vz

Vz = 0

Vz = VWOD

max

max

Vx = VWOD cos WOD

Vy = -VWOD sin WOD

CFD Generated Ship Airwakes

USN FFG

SFS

Flight Deck

Spot­#1

Spot­#2

Spot­#3

WOD

VWOD

HangarFace

150ft

50ft

30ft

30ft30ft

AreaofInterest

SFS Ship Airwakes

• Along-wind airwake velocities

WOD = 0°Recirculation

ZoneVWOD

50 kts

70 kts

60 kts

50 kts

WOD = 270°Flow

AccelerationZone

VWOD

50 kts

40 kts

40 kts

20 kts

2-D Aerodynamic Modeling

• Aerodynamics modeled with­ Nonlinear quasi-steady aerodynamics (Ref. Prouty & Critzos)» Aero forces dependent upon instantaneous values of ,,

­ Nonlinear time-domain unsteady aerodynamics (Ref. Leishman)» Aero forces dependent upon time history of , , » Model only validated for small and (< 25°) and M > 0.3

» Must switch to quasi-steady at high and (> 25°) and M < 0.1

• • •

• •

-3

-2

-1

0

1

2

3

0 45 90 135 180 225 270 315 360(deg)

+V

AngleofAttackConvention

cl

cd

cmc/4

Structural Modeling - Element

Weight

Aero

> > > > > > >• ••

CF

• FEM used to accommodate different hub geometries

- Articulated, hingeless, teetering and gimballed

• 11 degrees of freedom per element

- 4 flap, 4 lag, & 3 twist

• Distributed blade loads

- Inertial, Aerodynamic, Weight and Centrifugal Force» Inertial loads include rotor acceleration

vb

v’b

wb

w’b

b

m

va

v’a

wa

w’a

a

Structural Modeling - Blade

RotorShaft

Finite Element

(t)

Flap Hinge

Conditional Flap stop springs

K

Control StiffnessSpring

K

PitchBearing

LagHinge

Conditional Lag stop springs

K

• Articulated blade modeling

­ Require mechanisms to restrain flap (hinge) & lag (hinge) motion

» Stops simulated with conditional springs K and K

­ Flap stops extend/retract at a specified rotor speed

Structural Modeling - Rotor

• Articulated or hingeless rotors

• Teetering or gimballed rotors

2

1

3

Blade motions are uncoupled

1, 2 and 3 independent

1

2

M1

[Mrotor] = M2

M3

0 0

00

00

[Mrotor] = M2

0

0

Blade motions are kinematically coupled

1 = -2

M1

Introduction

Previous Research

Objectives

Approach

• Results­ Baseline rotor

­ Passive control of H-46 rotor

­ Feedback control of gimballed rotor

• Conclusions

Presentation Outline

Baseline Rotor System

• Representative of a “medium-sized” naval helicopter

­ Nb = 4 Articulated Blades

­ R = 25 ft

­ 0R = 750 ft/s

­ = 7.35

­ = 0.076

­ = 1.02/rev

­ = 0.30/rev

­ = 4.54/rev

­ FS = ±1º

­ LS = ±10º

0

20

40

60

80

100

0 5 10 15 20

(%NR)

Time(s)

FlapStopsRetract

Measured H-46

Baseline

Typical Engagement

• Linear airwake

­ VWOD = 60 knots

­ = 25%

• Largest wtip occur

< 25%NR

• Blade strikes flap stops repeatedly

• Majority of wtip is

elastic bending

­ rigid body wtip ±2%R

• Large in low even near blade tip -180

-90

0

90

180

0 2 4 6 8 10 12

50%R95%R

(deg)

Time(s)

-15

-10

-5

0

5

10

15

hinge

(deg) DS

FS

FlapStopsRetract

-30

-20

-10

0

10

20

30

wtip

(%R)

H-46 Tunnel

<25%NR

Typical Engagement

• Linear airwake

­ VWOD = 60 knots

­ = 25%

• Majority of vtip is

rigid body motion

• Blade strikes lag

stop repeatedly

• Largest torque due

to impacts

-20

-10

0

10

20

vtip

(%R)

-15

-10

-5

0

5

10

15

hinge

(deg)

LS

LS

-20000

-10000

0

10000

20000

0 2 4 6 8 10 12

Q(ft-lb)

Time (s)

-30

-20

-10

0

10

20

30

0 2 4 6 8 10

wtip

(%R)

H-46 Tunnel

Time (s)

Typical Wind Envelope

• Engagement wind envelope

­ Shows largest downward and upward wtip with VWOD and WOD

VWOD = 60 ktsWOD = 30°

Upward wtip

Downward wtip

Upwardwtip

Downward wtip

SFS Ship Airwake

• What effect does a “realistic” ship airwake have on rotor deflections?

SFS

Flight Deck

Spot­#1

Spot­#2

Spot­#3

WOD

VWOD

HangarFace

150ft

50ft

30ft

30ft30ft

AreaofInterest

Spot #1 Engagement Envelope

• Bow and port winds have largest wtip

• Stern and 330° winds have small wtip Spot #1(Closest to hangar)

Recirculation and downflow behind hangar face

RecirculationZone

VWOD

Downflow

RecirculationZone

VWOD

Recirculation zone pushed away from flight deck

VWOD

Upflow & FlowAcceleration Zone

Large upflow component on windward side of flight deck

VWOD

Upflow & FlowAcceleration Zone

Large upflow component over flight deck and over hangar face

VWOD

FlowDeceleration

Little upflow over stern and flow decelerates near hangar face

Effect of Deck Position

• Spots closer to hangar have larger wtip

• Largest wtip consistently in port winds

• wtip for Spot #1 are ~2wtip for Spot #3Spot #1

Spot #2

Spot #3Spot #1 Spot #2 Spot #3

Introduction

Previous Research

Objectives

Approach

• Results­ Baseline rotor

­ Passive control of H-46 rotor­ Flap Damping

­ Spoilers

­ Feedback control of gimballed rotor

• Conclusions

Presentation Outline

Objectives

•Develop unique “in-house” analysis code to:

- Increase physical understanding of engage/disengage behavior

- Accurately predict safe rotor engage/disengage envelopes

Safe Region

- Control rotor response to expand engage/disengage envelopes

Unsafe Region

wtip (%R)

• Hydraulic flap dampers were used on 1950’s era HUP-2

­ Dampers only active at low

­ Above preset dampers became inactive

• Use same technique on H-46 Sea Knight

­ Not necessarily traditional hydraulic damper - MR or ER?

­ Use of mast causes drag penalty in forward flight

Flap Damping on HUP-2

Blade

Hub

Mast

Counterweight

Spring

Damper

Flap Damper Sizing for H-46

• Examine “worst-case” scenario - Spot #1 Airwake

-30

-20

-10

0

10

20

30

40

0 1 2 3 4 5

Maxw

tip

(%R)

Flap Damper Strength (xC)

Minwtip

(%R) H-46Tunnel

C

H-46 flap stops set at ±1º

Flap damper has stroke of only

Majority of wtip is elastic

Flap damper has no effect with a

small stroke!

FS

Flap Damper Sizing for H-46

• Raise flap stop setting

­ Allows damper larger stroke

­ Keep droop stop setting at -1º No additional downward wtip

C

Raise flap stop setting

Flap damper has larger stroke

Flap damper has much large

effect!

FS

-30

-20

-10

0

10

20

30

40

0 2 4 6 8 10

Maxw

tip

(%R)

FS(deg)

Minwtip

(%R)

StandardConfiguration

C=4C

C=3C

C=5C

SFS Spot #1 Envelope

• Flap damper = 4C

• Flap stop = 6°

• Max wtip increased in

210°- 240° winds

+30%R to +34.8%R

• Min wtip decreased in

240°- 300° winds

-22.4%R to -14.8%R

• Min wtip not affected

in bow winds

Still -25.2%R

Maxwtip

Min wtip

Standard H-46 With Damper

Flap Damping in Bow Winds

• Blade does not lift

off DS until t = 5 sec

• Flap damper never

has a chance to

dissipate energy

• Summary:

­ Min wtip decreased

in most cases

­ FS must be raised

­ Max wtip increased

-30

-20

-10

0

10

20

30

wtip

(%R)

H-46 Tunnel

No Reduction

StandardConfiguration

FlapDamper

-2

-1

0

1

2

0 2 4 6 8 10

hinge

(deg)

Time(s)

StandardFS

DS

StandardConfiguration

FlapDamper

Introduction

Previous Research

Objectives

Approach

• Results­ Baseline rotor

­ Passive control of H-46 rotor­ Flap Damping

­ Spoilers

­ Feedback control of gimballed rotor

• Conclusions

Presentation Outline

• Examine reducing flapping by reducing excessive lift

• Leading-edge spoilers known to significantly reduce lift

Objectives

L

V

L

VWithout spoiler With spoiler

Leading-edge spoiler

(Ref. Brasseur)

Objectives

LowAF Low

CF

<25%SpoilersExtended

AppreciableCF

AppreciableAF

=25%SpoilersRetract High

AF

>25%SpoilersRetracted

HighCF

• Percentage of radius covered by spoilers?

• Will rotor torque increase due to spoiler drag?

• Spoilers are used only along partial-span

• Gated spoilers are used on blade upper and lower surfaces

• Spoilers only extended at low < 25%NR and retracted into blade section at high > 25%NR

Spoiler Coverage

H-46 Engagement

SFS Spot #1 Airwake

VWOD = 40 kts

WOD = 240°

• H-46 engagement with varying amounts of spoiler coverage

• Spoilers on outer 15%R (~3½ ft) are enough to reduce wtip

0

10

20

30

Max

wtip

(%R)

-30

-20

-10

0

0 10 15 25 50 75

xspoiler (%R)

Min

wtip

(%R)

X X % R

xspoiler

H-46 Tunnel

Example Engagement

SFS Spot #1 Airwake

(Worst Case Scenario)

VWODSpot #1

VWOD = 40 kts

WOD = 240°

Conclusions:

Min and Max wtip reduced

Max torque not affected

-30

-20

-10

0

10

20

30

wtip

(%R)

Spoilers DeployedH-46 Tunnel

38%Reduction 21%

Reduction

Standard Configuration

Spoilers

-15

-10

-5

0

5

10

15

hinge

(deg)

LS

LS

SpoilersDeployed

StandardConfiguration

Spoilers

-20000

-10000

0

10000

20000

0 2 4 6 8 10

Q(ft-lb)

Time (s)

Spoilers Deployed

Standard ConfigurationSpoilers

SFS Spot #1 Airwake Envelopes

Maxwtip

Min wtip

Standard H-46 With Spoilers• Max wtip decreased

in 210°- 270° winds

+30%R to +23%R

• Min wtip decreased

in 240°- 300° winds

-25.2%R to -17.5%R

• Min wtip decreased

in bow winds

-23%R to -18.5%R

• Conclusion:

­ Both Min and Max wtip reduced

Introduction

Previous Research

Objectives

Approach

• Results­ Baseline rotor

­ Passive control of H-46 rotor­ Flap Damping

­ Spoilers

­ Feedback control of gimballed rotor

• Conclusions

Presentation Outline

Motivation

Rubber Spring

Gimbal Restraint

• V-22 blades much shorter & stiffer than articulated blades

­ Rotor motion due to rigid body motion, not elastic bending

• V-22 utilizes active “flap limiter” to reduce flapping in FF

­ Feedback from gimbal motion to swashplate inputs

• Could flap limiter be used in engagement ops?

•Rigid blade structural model

­2 degrees of freedom - gimbal pitch (1c) and roll (1s)

•Linear quasi-steady aerodynamic model

­Lift >> Drag

z

x

y1C

1SKb

Structural & Aerodynamic Modeling

( )20 T P TL U U U

2

γ≅ θ −

­Control System Settings

Swashplate inputs

( ) )(kxsincos pp43

twis1ic1750 iβ−β+−θ+ψθ+ψθ+θ=θ

vBRKxu T1−−−=

( ) ( ) ( ) ( )ft T Tf f f 0

1 1J x t S t x t x Qx u Ru dt

2 2= + +∫

­ S(tf) = Final State Weight Q = State Weight R = Control Weight

• Use Matrix Ricatti Equations to find gain matrix K

( ) ( ) ( ){ } { }T

s1c175

T

s1c1s1c1 ux

tdutBxtAx

θθθ=ββββ=

++=&&

&

Disturbance d(t) due to:Airloads induced by ship

airwake effects

­ Equations are Linear Time Variant (LTV)

­ (t) and aerodynamic terms make pole placement ineffective

• Use LQR theory and define performance index J

Additional gain due to disturbance d(t)

Optimal Control Theory

• Cast equations of motion into state space form

vBRKxu T1demanded

−−−=

•Swashplate actuators typically have limits in magnitude and rate

oo

o

&

1010

5.7

7x

:Limits

c1

75

sin

maxac

<<−−>=Non-Rotating

Swashplate

RotatingSwashplate

Actuator#1

Actuator#2

Actuator#3

xac

Enforcecontrollimits

actualu dBuAxx ++= actual&

•Time integration with control system limits

Control System Limits

•Simulated V-22 engagement

­Vwod = 30 kts in Bow winds

­Uncontrolled case:

» 75 = 1c = 1s = 0

•Constant airwake distribution

­ = 25%

Conclusion:

max reduced by 50%

Min 75 limit reached

0

5

10

15

max

(deg)

GimbalRestraint

Uncontrolled

OptimalControl

-15

-10

-5

0

5

10

15

u(deg)

1c

75

1s

1cmax

1cmin

75min

-10

-5

0

5

10

0 2 4 6Time (s)

xac

(in/s)

x1x

2

x3

xac

max

xac

min

Vwod

Response in Constant Airwake

• Gain K and disturbance effect v are functions of the ship airwake

• Knowledge of the ship airwake is difficult to predict/measure

­ Ship anemometer reads relative wind speed and direction

­ Correlates to in-plane velocities Vx and Vy over flight deck

Anemometer

Vx and Vy may vary over the flight deck

Vz is unmeasured!

VWOD

Vz

Vy

Vx

WOD

Optimal Control Assumptions

Sub-Optimal Control

Conclusion:

Optimal gains max by 50%

Sub-optimal gains max by 35%

0

5

10

15

0 2 4 6

max

(deg)

Time(s)

GimbalRestraint

Uncontrolled

Sub-OptimalControl

OptimalControl

0

5

10

15

max

(deg)

GimbalRestraint

Uncontrolled

Sub-OptimalControl

OptimalControl

• V-22 Rotor Engagement

­ Vwod = 30 knots

­ Constant airwake

• Sub-Optimal Control

­ Vx and Vy known

­ Vz assumed = 0

• Optimal Control

­ (Best Case)

­ Vx, Vy and Vz known

•Anemometer measurement error

•Conclusion:

Moderate errors in anemometer reading change response by 10%

wodwodmeas

wodwodmeas VVV

Δ+=Δ+=

0

5

10

15

max

(deg)

ΔVwod=-10kts

ΔVwod=+10kts

Sub-OptimalControl

•Gains K and v calculated from

(incorrect) anemometer meas.

Error in Wind Velocity

0

5

10

15

0 2 4 6

max

(deg)

Time(s)

Δwod=-15deg

Δwod=+15deg

Sub-OptimalControl

Error in Wind Direction

Anemometererror

Robustness to Anemometer Error

Introduction

Previous Research

Objectives

Approach

Results

• Conclusions

Presentation Outline

Conclusions

• Developed transient elastic F-L-T analysis for E/D ops

­ Blade structure modeled with FEM

» Articulated, hingeless, teetering, or gimballed rotors

» Blade weight and acceleration included

­ Aerodynamics simulated with quasi-steady or unsteady models

­ Airwake modeled with simple types or from numerical predictions

­ Rotor motion time-integrated along specified (t) profile

• Investigated effect of “frigate-like” ship airwake

­ Blade wtip showed strong dependence on wind direction

» Winds off-bow had smallest wtip, winds over-port had largest wtip

­ Spots closer to hangar had larger deflections

Conclusions

• Investigated effect of flap damper for H-46

­ Raised flap stop setting to allow damper larger stroke

» Reduced downward wtip by 30%, but increased upward wtip by 20%

» Downward wtip not affected at all in some cases

• Investigated effect of leading-edge spoilers for H-46

­ Spoilers extend ( < 25%NR) and retract into blade ( > 25%NR)

­ Determined spoilers needed only on outer 15%R of blade

» Reduced upward and downward wtip by 20%

­ No significant increase in maximum rotor torque in any case

• Investigated control of gimballed rotors w/ LQR

­ Used feedback from gimbal motion to swashplate actuators

­ Resulting equations of motion were Linear Time Variant (LTV)

­ Enforced control system limits (magnitude and rate)

• LQR control method successful at reducing flapping

­ max 50% with full knowledge of ship airwake (Vx, Vy and Vz)

• Aero forces due to ship airwake contribute to control gains

­ max 35% with partial knowledge of ship airwake (Vx and Vy)

• Response insensitive to errors in anemometer reading

­ max changed ±10% with either ±10 knot or ±15° anemometer error

Conclusions

Acknowledgments

•Financial assistance

­National Rotorcraft Technology Center

»Technical Monitor Dr. Yung Yu

•Technical Assistance

­Dynamic Interface Group NAWC/AD Pax River, MD

»Mr. William Geyer, Mr. Kurt Long & Mr. Larry Trick

­Boeing Philadelphia

»Mr. David G. Miller